The 10r is not a ground lift.
It is an attenuation of the interference current brought about by the loop in the two input interconnects as described by D.Joffe (HBRR & HBRL).
No matter what language you've chosen for it, resistance on both ends of the too-small 150u has made a bad cap simulator, and no matter what any sort of botique cap is spent for it, none of them will work well at high loss circuit and at severely the wrong capacitance size (also high loss). That bit is absolutely guaranteed to ruin audio signal.
Edit: There are two problems, and I'd rather not hear either.
Edit: There are two problems, and I'd rather not hear either.
Last edited:
I think you can bypass the 10r with a 100n ceramic, thus removing my objections (and the relevant problems) at the cost of a few cents.
Otherwise. . .
I seriously doubt that there is any botique cap made available for any sum of money that can overcome the error of increasing the effective ESR of the inverting input coupler by 10. I'm not ever doing that.
I am not getting this. What are you talking about.
Go on, convince me that the 10r between the signal return and the power ground has the effect of increasing the esr of the DC blocking capacitor.
And plain technical english please, not your flowery language that this backwoods Scotsman cannot understand.
Howabout the Honey Badger schematic, which has the same resistor we're discussing as well as a wrong-way patch with a ringing bypass to the inverting input's coupling cap, instead of a non-ringing bypass of the resistor (specifically, that same resistor that we're discussing)? I propose that the extra 100n would work more suitably if relocated to the resistor.
The difference might be inaudibly tiny, had there not been gain applied.
In my understanding that resistor was never meant to break the current loop of the source. When the resistor is there, it's assumed that source shield is wired to a common ground point and then back to signal return on the PCB, and the resistor just bypasses that long wire run at RF. At audio frequencies the resistor is bypassed by several inches of wire, so doesn't really add any resistance to the input current loop.
We could be forgiven for thinking the resistor was really meant to break ground since there are so many schematics which omit this crucial information.
We could be forgiven for thinking the resistor was really meant to break ground since there are so many schematics which omit this crucial information.
I still don't understand. What is this "ringing bypass"?
What does it do to the ESR of the DC blocking capacitor?
I will work up a schematic example, soon but not immediately (well it needs to look elegant and also look like sense too; so, that's not very fast, for me).
As I stopgap, to buy some time until then, I will simply propose that RCR may behave differently than RC and not so easily interchangeable in some applications. Specifically, if we had actually wanted an RC, then we probably should have used it and not RCR instead (as well meaning as adding in an extra resistor may be, the cost may be a circuit that we didn't want). Well, that's the thing. Practical schematic examples to follow in some days (my attempt of that).
AndrewT, I need your help on that. On this Honey Badger Schematic, the part is R4 the honey badger <--link
I thought that the name of that resistor was "Small Signal Ground Lift" but, if that assumption was in error, then what I need is the correct name of that part.
I would like to, and actually need to, use the correct name of that part while discussing it. I have in mind a patch to reduce/remove the performance costs of using it. However, I need the correct name of it for that discussion.
I chose to use the Honey Badger schematic for this because it has the bypass cap for circumventing the costs of using that resistor, in the wrong place, on an otherwise excellent amplifier. Also chip amplifiers can use the same resistors and caps in their input areas. And, chip amplifiers have more compact layouts, so that if a bypass cap is not correctly sized nor correctly located close enough to the area it was supposed to help, then it goes badly.
The concern is if the treble losses are not patched, then chip amp builders will be much more likely to short the inverting input coupler and miss out on its safety, durability, realism and dynamics benefits. The reasons to use that coupling cap would make a very long list. And, neither you nor me would want that list to get shorter, especially because of a resistor that we don't need except during layout efforts. What I need from you is the right name of that resistor. After that I'm going to patch (obliterate) the costs of using it. I do need to name it right before then. Help?
The documentation is short for what I'm wanting say, so I really do need your help in saying it the right way. It will be terribly simple/easy after that.
Regarding post #630 by Mark Whitney,
I applied the RED alternative, but instead of 100 nF I use 150 nF... anyway, these two values can be exercised depending on one's available speaker (IMHO). I once simulated and applied also the values suggested by AndrewT, but it seemed 150 nF gives the bass response I was looking for.
At the (+) input of the LM1875, I use 3x22k (1% tol) in series.
R1 is changed to 10k log pot and the amp is as silent as without the pot. I tried 100k log pot, but the amp buzzing/humming when I touched the pot.
PSU uses 2x2200 uF per rail at PSU box (I separated the boxes for the amp and PSU) and 18CT Vac trafo, it gives me about +/- 24 vdc (depending on the mains voltage). The amp sounds just right to me given the speaker I am using as I am listening to various genres. Bass is there... Treble is okay... vocal is sufficient to me.
Thanks a lot for your kind help.
I applied the RED alternative, but instead of 100 nF I use 150 nF... anyway, these two values can be exercised depending on one's available speaker (IMHO). I once simulated and applied also the values suggested by AndrewT, but it seemed 150 nF gives the bass response I was looking for.
At the (+) input of the LM1875, I use 3x22k (1% tol) in series.
R1 is changed to 10k log pot and the amp is as silent as without the pot. I tried 100k log pot, but the amp buzzing/humming when I touched the pot.
PSU uses 2x2200 uF per rail at PSU box (I separated the boxes for the amp and PSU) and 18CT Vac trafo, it gives me about +/- 24 vdc (depending on the mains voltage). The amp sounds just right to me given the speaker I am using as I am listening to various genres. Bass is there... Treble is okay... vocal is sufficient to me.
Thanks a lot for your kind help.
referring to post630
22k & 1K (R3) are the normal feedback that determines the amplifier gain.
220uF (C2) is added to reduce the DC gain down to ~1
C2 should be selected to pass all the audio frequencies with minimal phase shift. Use C1 to control the LF passband of the amplifier.
R5 & C7 are the Cherry addition to adjust the LF phase shift.
He suggests that R5=2*the 22k = 44k as shown and that the 10uF*22k = R3*C2
If these two capacitors C7 & C2 are selected to pass all the audio signals, then there will be no Audio voltage across either of them. If the output offset is zero there will be no DC voltage across them. Measure the completed amplifier. Both AC and DC voltages should be <<10mV
The Cherry addition shows when you test with squarewaves and a scope. The extreme LF "squares up". This squaring of the squarewave shape indicates that the phase shift has been reduced.
The 100nF is an addition to the 10uF to give 10u1F
Changing this to 10u15F (an increase of 0.5%) should give no difference, if the capacitors do actually pass all the audio signal.
Did you select C2 to pass all the Audio Signal?
Did you measure your two resistors and your two capacitors to ensure you met the Cherry requirement that R3*C2 = Rupper*Ccherry?
22k & 1K (R3) are the normal feedback that determines the amplifier gain.
220uF (C2) is added to reduce the DC gain down to ~1
C2 should be selected to pass all the audio frequencies with minimal phase shift. Use C1 to control the LF passband of the amplifier.
R5 & C7 are the Cherry addition to adjust the LF phase shift.
He suggests that R5=2*the 22k = 44k as shown and that the 10uF*22k = R3*C2
If these two capacitors C7 & C2 are selected to pass all the audio signals, then there will be no Audio voltage across either of them. If the output offset is zero there will be no DC voltage across them. Measure the completed amplifier. Both AC and DC voltages should be <<10mV
The Cherry addition shows when you test with squarewaves and a scope. The extreme LF "squares up". This squaring of the squarewave shape indicates that the phase shift has been reduced.
The 100nF is an addition to the 10uF to give 10u1F
Changing this to 10u15F (an increase of 0.5%) should give no difference, if the capacitors do actually pass all the audio signal.
Did you select C2 to pass all the Audio Signal?
Did you measure your two resistors and your two capacitors to ensure you met the Cherry requirement that R3*C2 = Rupper*Ccherry?
Last edited:
Oh, I apologize...
What I meant was that I did not use 10 uF. I parallel 44k resistor with 150 nF cap. My simulation showed additional gain at some LFs... If I am not mistaken it was about the 50 or 60 Hz or something. I lost the file.... If only 10 uF was used, the bandpass was wide, but what I need was to increase gain for some lower frequencies and adding 150 nF or 100 nF helps me with that. So, the 150 nF cap was not additional to 10 uF, instead it replaces the 10 uF.
Having this configuration, bass response of my overall system is better compared to the datasheet circuit ( I also have this datasheet standard amp).
I hope this clarifies thing.
Thanks, AndrewT.
What I meant was that I did not use 10 uF. I parallel 44k resistor with 150 nF cap. My simulation showed additional gain at some LFs... If I am not mistaken it was about the 50 or 60 Hz or something. I lost the file.... If only 10 uF was used, the bandpass was wide, but what I need was to increase gain for some lower frequencies and adding 150 nF or 100 nF helps me with that. So, the 150 nF cap was not additional to 10 uF, instead it replaces the 10 uF.
Having this configuration, bass response of my overall system is better compared to the datasheet circuit ( I also have this datasheet standard amp).
I hope this clarifies thing.
Thanks, AndrewT.
I'm almost as old as you are, and therefore can't help much on the appropriate bypass cap practices. Sorry man, the bad news is that I have become old, may be, good news which is that I haven't become less stubborn. Well, there you have it. I sure would like to do better on that topic, but not today.
Last edited:
It will be useful to initially assume that return does go to ground plane most of the time. It is not normally useful to exceed that notion, unless you wanted to ask a better question.
I built an lm1875 kit from chipamp.com and I have a bit of an issue with noise. I have a cheap 50k pot attached to it for the moment which I intended to upgrade later. Turned all the way down the speakers are silent and as I turn it up there is a bit more noise. This isn't unexpected since my pot isn't great. However then I decided to use the preouts of my headphone amp and then just the 50k potentiometer all the way up because I figured that it was the same as having a 50kohm shunt resistor and shouldn't hurt anything. I still hear a hum even when the preamp is at the absolute minimum. When I have nothing plugged in the noise is even louder. I assumed that the noise was from the power transformer because it isn't shielded, however if that's the case shouldn't I still hear something if the potentiometer is turned all the way down?
I have included a schematic. My circuit matches all values except for R1 and R6 jumper. R1 actually wasn't soldered in so I just put a high value resistor in the schematic. Similarly, R6 is just a piece of wire but I put a really low value resistance in order to represent that. They are included in the schematic because they are on the board.
I have included a schematic. My circuit matches all values except for R1 and R6 jumper. R1 actually wasn't soldered in so I just put a high value resistor in the schematic. Similarly, R6 is just a piece of wire but I put a really low value resistance in order to represent that. They are included in the schematic because they are on the board.