It's interesting that you find it beneficial to split the supply to the BUF. I tried it once briefly, using discrete regulators (which sound so much better that any 3xx, 78xx it's not even funny) and could only hear very small imrovement. Other than that - lots of caps for no reason.
Separate regulators give you better variability in combination of opamps ( you can use low voltage ones ) , while buffer have better parametres by higher voltage. Lot of caps ? You can't use it . 😎
Upupa:
A nice circuit and PCB. Thanks for the christmas present!
But: is there no better choice, indeed, than 7815 or 7818?
They are very noisy.
At least, when I attach a LM3875 with a gain of 46 to a OPA627, driven by an 7818, I can hear the noise clearly through my >92dB/W speakers.
Franz
A nice circuit and PCB. Thanks for the christmas present!
But: is there no better choice, indeed, than 7815 or 7818?
They are very noisy.
At least, when I attach a LM3875 with a gain of 46 to a OPA627, driven by an 7818, I can hear the noise clearly through my >92dB/W speakers.
Franz
Franz, gain 46 is too high. Noise what you listen, comes from LM and from OPA, not from 7818. My circuit have SNR over 120 dB ( measured on AP 😉 ) , although there is this " tralala " regulators 😉 .
No, Upupa
Gain 46 is perfect for the LM3875 with tube buffer: absolutely no noise! Its definitively the OPA62 AND the 7818!
Franz
Gain 46 is perfect for the LM3875 with tube buffer: absolutely no noise! Its definitively the OPA62 AND the 7818!
Franz
Franz, noise on supply rail is supressed by PSSR, which is by normal opamp 70 - 100 dB. Which is voltage gain of 627 stage ? Or aren't oscilate ? Have you look at output with scope ? Cathode follower have gain less than one ( 0.95 - 0.98 typicaly ).
>Franz, noise on supply rail is supressed by PSSR, which is by normal opamp 70 - 100 dB.
Out of the datasheet, we know.
>Which is voltage gain of 627 stage ?
1
>Or aren't oscilate ?
No, Upupa!
>Have you look at output with scope ?
Yes, of course!
>Cathode follower have gain less than one ( 0.95 - 0.98 typicaly ).
Yes, out of old books.
Hey man: now it is me, to criticize something: your christmas present!
Please give us something better, than we had before!
Franz
Out of the datasheet, we know.
>Which is voltage gain of 627 stage ?
1
>Or aren't oscilate ?
No, Upupa!
>Have you look at output with scope ?
Yes, of course!
>Cathode follower have gain less than one ( 0.95 - 0.98 typicaly ).
Yes, out of old books.
Hey man: now it is me, to criticize something: your christmas present!
Please give us something better, than we had before!
Franz
Franz, I don't know, why you have on your amp noise with standard regulator. With this regulator I had measure on my amp SNR over 120 dB. Somewhere is error. Or are you claiming, that I'm liar ?
As long as you don't lie, no!
Don't forget to smile, Upupa!
O.K., there must be something somewhere else wrong, in my circuit
Franz
Franz, don't get me wrong here, but noise calculations are fairly straightforward and done daily by many EE:s in (do I dare to say a lot more demanding applications than audio)... 😉
Apart from the root cause of some noise types (flicker noise and burst "popcorn" noise) the theory behind noise is very well understood and the "black magic" factor is very small.
First, as Upupa said the noise from any LM78x regulators is pretty much negligible in these kind of circuits. Yes, there are many modern 3-terminal regulators who have a heck of a lot better line & load regulation and noise performance. For a single well-decoupled opamp with high PSRR the 78x is plenty enough. If you have measurements that tell otherwise I am looking forward to see them! 😉
Now let's look at the system noise in your case. First of all I agree with Upupa that a gain of 46 (ca 33 dB) is way to much for the LM3875 both from a noise perspective and otherwise (such as reduced PSRR and increased distortion).
Secondly, where do you have the volume control? Before or after the OPA627 buffer? Post the schematic of your amp and we can have a walk through it from a noise perspective! A good amp should be able to reach 120 dB SNR as Upupa said.
😀
Cheers
/Magnus
Apart from the root cause of some noise types (flicker noise and burst "popcorn" noise) the theory behind noise is very well understood and the "black magic" factor is very small.
First, as Upupa said the noise from any LM78x regulators is pretty much negligible in these kind of circuits. Yes, there are many modern 3-terminal regulators who have a heck of a lot better line & load regulation and noise performance. For a single well-decoupled opamp with high PSRR the 78x is plenty enough. If you have measurements that tell otherwise I am looking forward to see them! 😉
Now let's look at the system noise in your case. First of all I agree with Upupa that a gain of 46 (ca 33 dB) is way to much for the LM3875 both from a noise perspective and otherwise (such as reduced PSRR and increased distortion).
Secondly, where do you have the volume control? Before or after the OPA627 buffer? Post the schematic of your amp and we can have a walk through it from a noise perspective! A good amp should be able to reach 120 dB SNR as Upupa said.
😀
Cheers
/Magnus
O.K. I will investigated another day (my VBIGC is absolutely quite).
You may have a look at my workbench: I plan to finish this PL519 enhanced triode mode amp by wednesday:
About 15kg iron/cupper on the desk: mains trafo, two chokes and output trannies 😀
Franz
P.S.
The poti is sitting at the input, before the opa627, as I use the LM3875 in inverting mode.
You may have a look at my workbench: I plan to finish this PL519 enhanced triode mode amp by wednesday:
An externally hosted image should be here but it was not working when we last tested it.
About 15kg iron/cupper on the desk: mains trafo, two chokes and output trannies 😀
Franz
P.S.
The poti is sitting at the input, before the opa627, as I use the LM3875 in inverting mode.
possibly the reason lies in the output transformers of the valve buffer? is it possible your wires are picking up noise?
try this -- remove cables from opa amp and short the inputs to get a good 0V input. did the noise go away?
try this -- remove cables from opa amp and short the inputs to get a good 0V input. did the noise go away?
Ahh, pretty much what I suspected. From a noise point of view that is kind of the worst thing one could do. I reckon you are not familiar with the Friis noise formula for cascaded systems but the truth is all in there my friend!
Basically, to first attenuate a signal by a large amount and then amplify it is detrimental for noise performance. I think you can figure that out just by intuition. So the first thing is, if you don't need the full gain of 33 dB for the power amp, reduce it. I usually settle for a power amp gain of 20-26 dB since my speakers have a sensitivity of 100-110 dB. 26 dB is pretty normal for a commercial amp.
But the worst offender here is using the OPA627 simply as a unity gain buffer. You can think about it but I really can not give the full explanation without some math (the Friis formula). If you increase the gain of this stage to 10-20 dB and simultaneously reduce the gain of the power stage by the same amount you will get much better noise performance. Using low-value resistors in the feedback network reduces thermal noise contribution but don't go below 1k as the opamp will have a hard time driving those loads.
Using a FET-input amp as a buffer before the bipolar LM3875 is good from a noise perspective as you then drive the LM3875 from a low impedance source. (Would you have connected the LM3875 directly to the pot the noise level would have changed with the pot setting. See this thread/post for an explanation http://www.diyaudio.com/forums/showthread.php?s=&threadid=46575&perpage=10&pagenumber=2)
But you really should not have excessive amounts of noise even with the OPA627 as a unity gain buffer. Perhaps you have lots of interference type noise injected in your circuit or dirty supply rails. Measure everything until the source of the problem is nailed down. Or as Werner von Siemens would have put it - "Zu messen ist zu wissen".
/Magnus
Maybe you could help me a bit Swedish Chef, on a slighty different project.
I'm running a NI paralleled 4780 chip amp with a gain of 21. This is being fed by a PGA2310 volume control chip that's master volume, and the PGA2310 is buffered by a OPA2134 with a gain of 2 (10k||10k, NI). Would that be fairly efficient noise wise, or would there be other configurations I could look at to test that might give superior performance?
I'm running a NI paralleled 4780 chip amp with a gain of 21. This is being fed by a PGA2310 volume control chip that's master volume, and the PGA2310 is buffered by a OPA2134 with a gain of 2 (10k||10k, NI). Would that be fairly efficient noise wise, or would there be other configurations I could look at to test that might give superior performance?
Magnus
Thanks, I appreciate your input, as well as input from Upupa! I am always listening to amps AND to EE's 🙂
Sorry, when I was a little bit "agressive" today...
I will print the posting from Magnus out and one day modify / examine this amp.
Franz
P.S.
I try to translate: "who measures, is measuring bulls***"
Thanks, I appreciate your input, as well as input from Upupa! I am always listening to amps AND to EE's 🙂
Sorry, when I was a little bit "agressive" today...
I will print the posting from Magnus out and one day modify / examine this amp.
Franz
P.S.
Sometimes this is more true: "Wer misst, misst Mist!"
I try to translate: "who measures, is measuring bulls***"
- Status
- Not open for further replies.