If the chipamps scare you, the LME49810 will not make your life any easier. The LME498xx series is awesome, though. I have an LME49811 driving a pair of STD03s running at 0.0002 % THD (1 kHz, 100 W, 8 ohm) if I recall the data correctly. Here's the build thread: Yet Another LME49810 + STD03 Amp.
How so? It's as close to a plug&play solution as you'll get. You "just" have to get the "plug" part right... 🙂 That's true for the LME498xx as well.
~Tom
Last edited:
Basically, I'm not a designer. I am above average practically speaking (for a uni student, which doesn't say much, i know)
I built P3 of the Elliot site, messed up the miller cap by a factor of ten, sorted it. Used TIP bourns devices rather than the 2N versions. Paralleled OP devices. Trimmed the bias (techy term unsure of) to minimise crossover glitching. Checked for stability. My only issue was somewhat naive PCB design, which left a slightly higher than desired white noise floor. Otherwise i did ok.
As for these IC amps...well they're easy in comparison (supposedly) but some of the more techy comments...well I got the impression (perhaps in error) that id get a lesser result than from my P3 effort.
If I was more confident to design discrete i probably would. Ive got final year project coming up, so perhaps its worth a stab �� ive already got the active filter stage done anyway.
I built P3 of the Elliot site, messed up the miller cap by a factor of ten, sorted it. Used TIP bourns devices rather than the 2N versions. Paralleled OP devices. Trimmed the bias (techy term unsure of) to minimise crossover glitching. Checked for stability. My only issue was somewhat naive PCB design, which left a slightly higher than desired white noise floor. Otherwise i did ok.
As for these IC amps...well they're easy in comparison (supposedly) but some of the more techy comments...well I got the impression (perhaps in error) that id get a lesser result than from my P3 effort.
If I was more confident to design discrete i probably would. Ive got final year project coming up, so perhaps its worth a stab �� ive already got the active filter stage done anyway.
Tom,
No problem, English is not my first language - and I find it hard to express my thoughts into words at the best of times.
After playing with the Audiolab MPWR and tweaking its design I've now had my fun and passed it on - I'll find out this weekend if its new owner is happy with its SQ.
Yes - the Negative output stage looks very dodgy in the Sims - I to had many problems with upper half being stable - but the lower breaking into oscillation... Non complementary output stages are never well behaved...
On the Amp I modified I added a CCS to the negative rail to offset the output stage crossover point away from the Zero crossing point - Constrained by the heatsink size I was able to sink about 200mA (with 37V rails), which resulted in the heatsinks temperature idling at about 55Deg.
This cleaned up the crossover distortion (leaving only odd order components 🙄 ) - as did reconfiguring the design for Shunt operation.
The CCS had to be sequenced with the LM3886 Mute pin to insure no Turn Off Currents (DC) as the PSU rail voltages reduced to Zero...
I found that the LM3886 frontend is VERY sensitive to demodulating GSM TDMA noise - forcing me to add 47pF to Ground on each input - slightly reducing the Phase margin - but atleast offering a level of attenuation to RF energy.
I'm very surprised that even with the 1.1K Emitter degeneration resistors on the Diff Pair that the front end is so sensitive to RF demodulation - in fact I wonder if the issue might be with the Emitter Follower Buffers in font of the Diff pair - otherwise the Diff pair Bipolar transistors must be really crappy (I also notice that Diff Pair is not Cascaded which will not aid HF linearity).
Sound quality - well they have very "full sound", making you tap your feet - no hint of nasty Brightness... (which is a pleasant surprise) - but they miss out on ultimate definition... Setting the PM to 82 Degs resulted in a very good L/R sound stage... All things considered, decent performance for so little - there sound "Balance" was better then the Audiolab Monoblocks I use in my Lab... Now they have gone, I miss there "non anaemic", bubbly, foot tapping sound... 🙂
Last edited:
Not exactly surprising given that the LM3886 is from the early 1990ies if I recall correctly. It wasn't until the 2000s that "EMI hardened" op-amps appeared and I'm not sure they've fully caught on.
On any wide band audio amp, an RF filter is definitely needed on the input.
~Tom
A couple of FFT plots:-
LM3886 Non Inverting 1W 8R (THD 0.004%)
LM3886 Inverting & 200mA ClassA CCS Sink (1W 8R, 0.0004% THD).
Shame about the Odd order crossover components...
LM3886 Non Inverting 1W 8R (THD 0.004%)
An externally hosted image should be here but it was not working when we last tested it.
LM3886 Inverting & 200mA ClassA CCS Sink (1W 8R, 0.0004% THD).
An externally hosted image should be here but it was not working when we last tested it.
Shame about the Odd order crossover components...
Last edited:
An externally hosted image should be here but it was not working when we last tested it.
1W 8R Inverting - Sans CCS
The addition of the CCS reduces the higher Even order components
Last edited:
Its such a shame the Crossover distortion is so high, otherwise the LM3886 would have made a decent general purpose Amp.
That was fast...really fast...thank you John!
Indeed...Have you had a chance to experiment with TDA7293/4?
Indeed...Have you had a chance to experiment with TDA7293/4?
I used the TDA7293/4 in our designs for Peachtree Audio - I prefere the sound of the TDA's to the LM's... Both are OK considering there simplicity - the LM's have tighter more fun Bass, while the TDA's have a more open "clearer" sound...
Last edited:
No Problem - I had the results saved at various stages as I worked on the unit - I've not heard from its new owner who was going to listen to it over the weekend - slightly worried, but hopefully I'll receive some good news 🙂
In order to extract maximum out of TDA7293/4 one needs to dispenses with the bootstrap cap and use a separate power supply for the driver stage. Everything begins here with almighty ilimzn...
http://www.diyaudio.com/forums/chip-amps/3857-tda7293v-2.html#post873005
http://www.diyaudio.com/forums/chip-amps/3857-tda7293v-2.html#post873005
I think you'll be very interested in US4739280 patent from Masao Noro (also a guru) ... a long time ago I made it with the Philips NE5532 opamp out of curiosity. Works perfectly.
I like to clarify that the earlier "Inverting mode" FFT plots the LM3886's overall gain has been reduced by 6dB to 25.2dB (hence the 6dB reduction in FFT noise floor over the Non inverting mode FFT).
the advantage of a stupidly obvious expired patent is that at least you now have it as incontrovertible "prior art"
unfortunately doesn't necessarily prevent a new patent issuing on another blindingly obvious variation of the same idea
unfortunately doesn't necessarily prevent a new patent issuing on another blindingly obvious variation of the same idea
Poor sound quality from LM3886
The bypass capacitors are too small as mentioned by someone else I would recommend not 1000 but 10000 uf capacictors to lower the impedance of the power supply. They should be kept very close to the IC. After the 10000 uf capacitors you would need 0.1 uf disc ceramic capacitors very close to the power supply pins to stop any rf interference injecting into those pins.You have not mentioned what value pot you are using for the volume control? Use a 10K pot as using higher values could roll off the high frequency response.Also you have not mentioned the value of the inductor which is marked as L this should be 0.7uH. I would also swap the 20K Rf and Crf 50pF which are in series the other way round as shown in the manufacturers datasheet. Also since since you are using the zobel network make sure you don't already have a zobel network inside the speakers. Listen to the sound without the zobel network and compare it with the zobel network in place. Also try the amplifier with different speakers.
The bypass capacitors are too small as mentioned by someone else I would recommend not 1000 but 10000 uf capacictors to lower the impedance of the power supply. They should be kept very close to the IC. After the 10000 uf capacitors you would need 0.1 uf disc ceramic capacitors very close to the power supply pins to stop any rf interference injecting into those pins.You have not mentioned what value pot you are using for the volume control? Use a 10K pot as using higher values could roll off the high frequency response.Also you have not mentioned the value of the inductor which is marked as L this should be 0.7uH. I would also swap the 20K Rf and Crf 50pF which are in series the other way round as shown in the manufacturers datasheet. Also since since you are using the zobel network make sure you don't already have a zobel network inside the speakers. Listen to the sound without the zobel network and compare it with the zobel network in place. Also try the amplifier with different speakers.
An observation that this Zobel may benefit from resistance being 10 ohms, rather than 3.3 ohms, and looks very similar despite different values to the zobel and inductor parallel resistor found in a quad 303.🙂
Cheers / Chris
Chris,
I agree with you that one would normally see 10R used in the output Zobel - but I found via simulation results that using 10R resulted in peaking - which could be resolved by using a lower value resistor - somewhere around 3 Ohms.
The datasheet shows a value of 2.7 ohm being used but the optimum value would be different for different speakers. The aim of the Zobel is to make the amplifier see a consistent load impedance across the whole range of frequencies the particular speaker (usually the bass speaker) is designed to handle. Personally I have used a 4.7 Ohm (5 watt) resistor with good results.
- Status
- Not open for further replies.