Good mornign FF
Thanks you for your explanation. yes i agree the caps a not so expensive...but with my housing i will look for ordinary caps 35V/3300µ -4700µ.....
pop off
i try it several times it is the amp itself. the dac is still running without signal and if i switch off (actually no switch installed at my mock up amp- just pull out the wall socket) - then i get a pop off on both channels
so pop off is a topic i have to solve - the chip has no mute pin to add some circuit there....actually i pull 1 speaker connector from the amp and then switch off
chris
Thanks you for your explanation. yes i agree the caps a not so expensive...but with my housing i will look for ordinary caps 35V/3300µ -4700µ.....
pop off
i try it several times it is the amp itself. the dac is still running without signal and if i switch off (actually no switch installed at my mock up amp- just pull out the wall socket) - then i get a pop off on both channels
so pop off is a topic i have to solve - the chip has no mute pin to add some circuit there....actually i pull 1 speaker connector from the amp and then switch off
chris
Hi Rabbitz
yes you are absolutly right, but with my experience at this weekend sound wise i like the "original" more
next step is to build a bigger cap bank...as i read this chip amps need this .. i do not expect that the amps sound so good that i push it to its limit
2nd step change the input caps at the kit to foil and the feedback cap.
chris
ESP has 2 x 4700uF for Project 72 and a minimum of 2 x 2200uF is recommended.
I don't know if Europe has the same deal but Digikey has free shipping of orders over AU$60.
I use a Kemet MMK on the input and I believe the feedback cap should be of good audio quality like Nichicon ES for bipolar or Nichicon KZ if not. IIRC, I used Nichicon ES 22uF 25V initially as I couldn't get
a higher value to fit. I ended up with a Nichicon FG 47uF 25V as my KZ would not fit.
I don't know if Europe has the same deal but Digikey has free shipping of orders over AU$60.
I use a Kemet MMK on the input and I believe the feedback cap should be of good audio quality like Nichicon ES for bipolar or Nichicon KZ if not. IIRC, I used Nichicon ES 22uF 25V initially as I couldn't get
a higher value to fit. I ended up with a Nichicon FG 47uF 25V as my KZ would not fit.
hi rabbitz
about supply caps
for normal listening the cap bank i am actually useing is fine but to try out if bigger caps make a better/lower ripple and therefore better SQ at the foggy background is interesting for me and i do not try 4 ohm load/speaker -my KEF Q100 are bastards
therefore i will check the power or current capability of my supply. the voltage sag measrued with my DMM is not really fast-on the scope you see that its more down that i see on DMMso before i make my "final LM1875 amp" i want to do some exersises / SQ check.....i am still learning
caps i agree with you the green muse nichicon ES are very good -film caps i have to look into my box -just Wima ... I will see
IIRC, I used Nichicon ES 22uF 25V initially as I couldn't get
a higher value to fit. I ended up with a Nichicon FG 47uF 25V as my KZ would not fit
which one is that? IIRC = iinfinite impulse response caps? the 330pF?
or you mean the Zobel cap? at the schematic with the 1R resistor value is 220nF?
chris
Attachments
Hi Rabbitz
about the NFB cap i found something from a very good engineer an advise
LM1875 power supply
post #10
the 22uF NP in the feedback loop is far too small.
Try >=100uF. I would suggest 220uF.
post #16
with the low value feedback capacitor (DC blocker), that capacitor forms a high pass filter to roll off the bass response of the amplifier. This is a bad way to band limit an amplifier. Most reliable authors/designers warn against this practice. (it develops AC voltage across the electrolytic capacitor).
The DC blocker must be made large enough that any operating voltage across it is negligible.
post 18
The NFB DC blocker generally has a small DC voltage across it. A polar electrolytic must be oriented to match that small voltage offset.
If the amplifier fails catastrophically then a much larger offset can be applied to that electrolytic. This can be guarded against by adding inverse parallel connected diodes across the DC blocker cap.
If the electrolytic is sized appropriately then the AC voltage across it is very low. For the electrolytic to work as designed it does not hinder it's operation if the small DC + the small AC signal across it can have +ve or -ve values. But the instantaneous sum of voltage values must be kept small.
That is where the Passive Input Filter and the NFB DC blocker ratio of values comes into play.
On the other hand, where a Manufacturer fiddles with that cap value ratio to "adopt an in-house sound", then there is the possibility that the AC signal is no longer small. In this situation a non-polar electrolytic may be required to give reliable long term stability in performance.
post #23
The 22uF non-polar electrolytic can be replaced with a 150uF or 220uF polar electrolytic. A low voltage rating is OK here. Try 16Vdc or even 10Vdc. It should be quite small and cheap. But add those protective diodes across the cap pins.
chris
about the NFB cap i found something from a very good engineer an advise
LM1875 power supply
post #10
the 22uF NP in the feedback loop is far too small.
Try >=100uF. I would suggest 220uF.
post #16
with the low value feedback capacitor (DC blocker), that capacitor forms a high pass filter to roll off the bass response of the amplifier. This is a bad way to band limit an amplifier. Most reliable authors/designers warn against this practice. (it develops AC voltage across the electrolytic capacitor).
The DC blocker must be made large enough that any operating voltage across it is negligible.
post 18
The NFB DC blocker generally has a small DC voltage across it. A polar electrolytic must be oriented to match that small voltage offset.
If the amplifier fails catastrophically then a much larger offset can be applied to that electrolytic. This can be guarded against by adding inverse parallel connected diodes across the DC blocker cap.
If the electrolytic is sized appropriately then the AC voltage across it is very low. For the electrolytic to work as designed it does not hinder it's operation if the small DC + the small AC signal across it can have +ve or -ve values. But the instantaneous sum of voltage values must be kept small.
That is where the Passive Input Filter and the NFB DC blocker ratio of values comes into play.
On the other hand, where a Manufacturer fiddles with that cap value ratio to "adopt an in-house sound", then there is the possibility that the AC signal is no longer small. In this situation a non-polar electrolytic may be required to give reliable long term stability in performance.
post #23
The 22uF non-polar electrolytic can be replaced with a 150uF or 220uF polar electrolytic. A low voltage rating is OK here. Try 16Vdc or even 10Vdc. It should be quite small and cheap. But add those protective diodes across the cap pins.
chris
Chris
I refer to the 22uF 16V NP in the schematic.
Yes, I've seen up to 220uF in this location but it's hard to get a quality larger cap in there as 8mm diameter is the largest that will fit with sitting on the PCB.
Correct, it is a good video for understanding the problematic.
For each channel and with the supply voltage he uses, more than 4700uF adds little to the level where the output stage is limited by the momentary supply voltage. For two channels and the same voltage, it becomes 2x4700uF=10000uF on each rail.
My impression was that you wanted to investigate the influence of rail ripple on the "foggy" sound you hear, without (below) the output being limited by the instantaneous voltage on the rails.
The off pop problem is solved by a relay that only switches the speakers in when the rail voltage is above 70% of the operational voltage. Class AB chips provide the basic amplifier functionality, most often not much service functions.
Hi FF
Thanks for your time and explanation.
Actually i add at every rail (4) 3300µF directly at the +V-gnd-V terminal....i hate this terminals !!!! so fiddly...
I here since 3 hours now... what i claimed as foggy is the "background" music - its the not correct or pronounced instruments and vocals and the fog around them. + the bloated bass - i really not request a lot of watts = power test as i posted before with the "flamenco"
so per rail and amp i have 2200µ+1000µ+3300µ = 6500µ - so per rail about 13mF
sound much better now...so its time to change the small caps a the amp board...
is the higher ripple infect the 1875 chip so hard?? the PSSR is according to the datasheet 52dB as tested. so if i understand correct as more ripple at the supply voltage i get more "noise" at the output of the amp? and this is the reason if i minimize the ripple at the supply i get a "cleaner" and more silent amp?
if 0dB is 1 and 50dB is 0,0032...how does the gain of 27dB comes into play here?
sorry for the nooby question
pop on is now additionally to the pop off...is an extra task...
chris
Thanks for your time and explanation.
Actually i add at every rail (4) 3300µF directly at the +V-gnd-V terminal....i hate this terminals !!!! so fiddly...
I here since 3 hours now... what i claimed as foggy is the "background" music - its the not correct or pronounced instruments and vocals and the fog around them. + the bloated bass - i really not request a lot of watts = power test as i posted before with the "flamenco"
so per rail and amp i have 2200µ+1000µ+3300µ = 6500µ - so per rail about 13mF
sound much better now...so its time to change the small caps a the amp board...
is the higher ripple infect the 1875 chip so hard?? the PSSR is according to the datasheet 52dB as tested. so if i understand correct as more ripple at the supply voltage i get more "noise" at the output of the amp? and this is the reason if i minimize the ripple at the supply i get a "cleaner" and more silent amp?
if 0dB is 1 and 50dB is 0,0032...how does the gain of 27dB comes into play here?
sorry for the nooby question
pop on is now additionally to the pop off...is an extra task...
chris
Last edited:
I'm running this chip w/o the neg fdbk capacitor allowing response all the way to DC. So far, no problem meow. I would think C 2.2uf on the input would block any DC getting into it for further amplification.
That cap I can understand in the single supply version, but why its necessity with +/- supplies I fail. Personally, I dont want any electrolytic cap in my signal path... That's just purely from audiophile experience.
All caps have a "voice" and there can be practitioners who like the sound of ...polarized electrolytics conducting AC. If that's what someone likes, no point in challenging, but methinks the sound of a "wire" is better, for obvious reasons.
That cap I can understand in the single supply version, but why its necessity with +/- supplies I fail. Personally, I dont want any electrolytic cap in my signal path... That's just purely from audiophile experience.
All caps have a "voice" and there can be practitioners who like the sound of ...polarized electrolytics conducting AC. If that's what someone likes, no point in challenging, but methinks the sound of a "wire" is better, for obvious reasons.
Last edited:
Hi FF
Thanks for your time and explanation.
Actually i add at every rail (4) 3300µF directly at the +V-gnd-V terminal....i hate this terminals !!!! so fiddly...
I here since 3 hours now... what i claimed as foggy is the "background" music - its the not correct or pronounced instruments and vocals and the fog around them. + the bloated bass - i really not request a lot of watts = power test as i posted before with the "flamenco"
so per rail and amp i have 2200µ+1000µ+3300µ = 6500µ - so per rail about 13mF
sound much better now...so its time to change the small caps a the amp board...
is the higher ripple infect the 1875 chip so hard?? the PSSR is according to the datasheet 52dB as tested. so if i understand correct as more ripple at the supply voltage i get more "noise" at the output of the amp? and this is the reason if i minimize the ripple at the supply i get a "cleaner" and more silent amp?
if 0dB is 1 and 50dB is 0,0032...how does the gain of 27dB comes into play here?
sorry for the nooby question
pop on is now additionally to the pop off...is an extra task...
chris
Hi Chris,
One thing is what we can estimate in an engineering manner from the ripple level and the PSRR. Another is what skilled audiophiles like you can actually hear. It can be difficult to relate the two. I appreciate very much your very thorough listening tests because that's where I am miles behind you. As an experienced engineer I can often conclude logically on clear technical improvements but their real influence on SQ is difficult for me. I just learned about "bootstrap snubbers" and their SQ effect should exceed what I would expect from an engineering point of view.
Bonne nuit..........
Last edited:
Thanks FF
I do not think that nobody is "able" to hear that - you have to spend time and try out. the age is a topic but i think famous music conductors hear a lot of thing from their orchestra and their ar mostly not young.
so acccording to the post 195 we have after the CRC cap bank per rail a 220mV ripple with 200hz.
I found a formula :
C = 1/( V^2) * f ---> 1/(0,22^2) * 200 = 4132pF = 4,1µF
R should be something 1 - 10R
chris
I do not think that nobody is "able" to hear that - you have to spend time and try out. the age is a topic but i think famous music conductors hear a lot of thing from their orchestra and their ar mostly not young.
so acccording to the post 195 we have after the CRC cap bank per rail a 220mV ripple with 200hz.
I found a formula :
C = 1/( V^2) * f ---> 1/(0,22^2) * 200 = 4132pF = 4,1µF
R should be something 1 - 10R
chris
Thanks FF
I do not think that nobody is "able" to hear that - you have to spend time and try out. the age is a topic but i think famous music conductors hear a lot of thing from their orchestra and their ar mostly not young.
so acccording to the post 195 we have after the CRC cap bank per rail a 220mV ripple with 200hz.
I found a formula :
C = 1/( V^2) * f ---> 1/(0,22^2) * 200 = 4132pF = 4,1µF
R should be something 1 - 10R
chris
Hi Chris, are you about to construct a snubber to remove some of the higher frequency noise? I am not sure if you will succeed. It may be sampling noise from your scope or it may be too low impedance for you to remove it.
You say 200Hz. Isn't it 100Hz?
How does a CRC-bank change the ripple frequency from 100Hz to 220Hz? My guess is that your trigger level or scope time-base are set so that the scope shows 220Hz with a large spread.
Chris, look at the horizontal settings: first photo 10ms per division, second and third photos 5ms per division. First photo one division per signal period, second and third photos two divisions per signal period. The signal period is 10ms for all three photos, hence 100Hz.
Last edited: