Hi,
Just a quick question on GC's and zobel networks.
I have an Audiosector LM3875 GC, running in dual mono configuration, driving 8 ohm speakers. I would like to try out a Zobel network, mainly for peace of mind.
I was wondering what the stock values are (or should be) for R and C, and what type of capacitor is the best to use?
Chris.
Just a quick question on GC's and zobel networks.
I have an Audiosector LM3875 GC, running in dual mono configuration, driving 8 ohm speakers. I would like to try out a Zobel network, mainly for peace of mind.
I was wondering what the stock values are (or should be) for R and C, and what type of capacitor is the best to use?
Chris.
0.1uF/2.7ohm? What effect will these values have on the sound, and will a ceramic capacitor (I have some 0.1uF Kemet's in my spares box) be Ok in this application?
Regards,
Chris.
Regards,
Chris.
Yes, those are correct values, and they also have been specified here: http://audiosector.com/lm4780 amp.pdf
Use metalized film cap, similar to this: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=BC2054-ND
Use metalized film cap, similar to this: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=BC2054-ND
Those parts will add coloration to the sound, I never tried ceramic in this application, so can't comment, but generally film would be better.
As well as giving me peace of mind, I am hoping the zobel network will solve a minor issue I have. Occasionally I get interference (using FM receiver) which causes a humming noise at the speakers. This noise remains present even after the source attenuator (50k Alps blue, dual log with a 47k metal film shunt) is turned down to minimum, i.e. input shorted to earth.
Chris.
Chris.
It could be Zobel, or maybe you need this mod: http://www.diyaudio.com/forums/showpost.php?p=1516522&postcount=51
hello.
for the reason of electrical safety.a cap rated 100v dc should withstand 63v ac .......
greetings
I have fitted a zobel to the output using the values 0.1uf/10R (0.6W). I used a metal film resistor (for temp stability) and a good quality ceramic capacitor. I went with these values on recommendation from someone else. He suggested using 10R to bring the 3dBpoint to around 100KHz, still way above audibility. Also, if the amp does oscillate at a higher frequency (588KHz using 0.1uF/1R) the resistor could get very hot. The zobel should not affect the circuit until way above audible frequencies anyway, so the quality of the capacitor should not matter, just the value.
Upon first listening (low level due to the time of night) I could not detect any colouration in the sound, and the treble was not rolled off at all. I'll have a proper listen later at a decent volume. I'm happier that my amp is now protected from possible damage due to HF oscillations. A cheap mod and only 20 minutes work.
Chris.
Upon first listening (low level due to the time of night) I could not detect any colouration in the sound, and the treble was not rolled off at all. I'll have a proper listen later at a decent volume. I'm happier that my amp is now protected from possible damage due to HF oscillations. A cheap mod and only 20 minutes work.
Chris.
Ceramic capacitors are not recommended for anything connected to the audio path because their value changes quite drastically with vibration. That said, their small size makes them a lot easier to fit.
The ceramic capacitors in question are Kemet CKR06, MIL Spec (MIL-PRF-39014). Also, my amp is well isolated from vibration, so I am not worried about the caps. The values shouldn't change any more than the specified 10% tolerance anyway. The zobel networks are in the audio path (although not directly), however they don't actually affect the sound until way above audible frequencies.
I had a good listen to the system this afternoon. I could not detect any colouration to the sound and the treble did not seem rolled off at all. If anything, the bass seemed to have a bit more clout. I can't see a reason why this would be. I'll have to give the system another listen tomorrow as this may be down to external factors (mains, background noise etc).
Overall, I am very happy with my 20 minutes work.😀
Chris.
I had a good listen to the system this afternoon. I could not detect any colouration to the sound and the treble did not seem rolled off at all. If anything, the bass seemed to have a bit more clout. I can't see a reason why this would be. I'll have to give the system another listen tomorrow as this may be down to external factors (mains, background noise etc).
Overall, I am very happy with my 20 minutes work.😀
Chris.
higher K ceramic types have large negative coefficient of C vs V - I've been forced to spec 50 V caps in a +/- 5 V system to keep delta C < 10%
there is also an interesting design issue - the R should have power rating sized for full scale output at the max "audio" frequency you can pass - allowing for the C impedance division - carbon composition are often used in this position for the high surge rating and low inductance
there is also an interesting design issue - the R should have power rating sized for full scale output at the max "audio" frequency you can pass - allowing for the C impedance division - carbon composition are often used in this position for the high surge rating and low inductance
Hi Chris,
I have also been wondering about the resistor value for the Zobel network. I have seen proven designs using the 10 ohm value, but the National datasheets give values of 2.7 to 4.7 ohms depending on the chip. Could you elaborate on your reasons for choosing 10 ohm over the 2.7 as recommended by Peter?
Thanks
Chris DC
I have also been wondering about the resistor value for the Zobel network. I have seen proven designs using the 10 ohm value, but the National datasheets give values of 2.7 to 4.7 ohms depending on the chip. Could you elaborate on your reasons for choosing 10 ohm over the 2.7 as recommended by Peter?
Thanks
Chris DC
The reason I was given for 10 ohms and a capacitance of 0.1uF (to quote directly) was that it will have a 3dB point around the 100kHz level, still way above audio frequencies, but at least it will be doing something before it gets into RF.
Finally, a resistance of 2.7 ohms means that the amp will be seeing that resistance at very high frequencies, and if there are any spurious oscillations or rf getting into the amp, it will be driving a relatively low impedance, and may get hot.
I have been running the amp with these values for over a week now and had no issues. It still sounds great.
Regards,
Chris.
Hi Chris,
So with a 2.7e resistor the 3dB point is at a much higher frequency? And there is a heat penalty in case of trouble with RF? I am keen to avoid unnecessary heat because the amp will eventually end up in a rack with other equipment and ventilation may be limited.
I'm glad you are enjoying your amp. More about 10e resistors in my reply to Peter - read on.
Regards
Chris DC
So with a 2.7e resistor the 3dB point is at a much higher frequency? And there is a heat penalty in case of trouble with RF? I am keen to avoid unnecessary heat because the amp will eventually end up in a rack with other equipment and ventilation may be limited.
I'm glad you are enjoying your amp. More about 10e resistors in my reply to Peter - read on.
Regards
Chris DC
Hi Peter,
There are firms in the US that specialise in manufacturing equipment for the custom install market, basically distributed sound for peoples homes and offices. Some of this equipment uses chip amps like the 3886 and the smaller 2 channel 1876. These amplifiers often have to deal with very long cable runs (50 meters and more) and a wide variety of speaker loads, sometimes with a transformer coupled volume control (on the speaker line) thrown in. At least one of these brands uses a 10e resistor in series with a 22n 100v Mylar cap. As these products have proven themselves in very hostile situations over many, many years, I cant help but wonder why their designers opted for the 10e value. After all, they would have had access to the same data sheets that we do. I should add that all these designs use the inductor // resistor combination as per note 13 on the datasheet, as well as the Zobel network. I based my 3886 on the national datasheet and calculated my own values for Rr, Ri and Ci, which as it turned out are very different from the "default" values shown on the datasheet. I did the paperwork for my application about 2 years ago so I don't remember what the design parameters were anymore but basically I wanted to improve on what they (imported equipment) had, both sonically and power wise. Do you have a technical explanation why 10e is better and what effect a 22n cap will have instead of the 100n?
Regards
Chris DC
There are firms in the US that specialise in manufacturing equipment for the custom install market, basically distributed sound for peoples homes and offices. Some of this equipment uses chip amps like the 3886 and the smaller 2 channel 1876. These amplifiers often have to deal with very long cable runs (50 meters and more) and a wide variety of speaker loads, sometimes with a transformer coupled volume control (on the speaker line) thrown in. At least one of these brands uses a 10e resistor in series with a 22n 100v Mylar cap. As these products have proven themselves in very hostile situations over many, many years, I cant help but wonder why their designers opted for the 10e value. After all, they would have had access to the same data sheets that we do. I should add that all these designs use the inductor // resistor combination as per note 13 on the datasheet, as well as the Zobel network. I based my 3886 on the national datasheet and calculated my own values for Rr, Ri and Ci, which as it turned out are very different from the "default" values shown on the datasheet. I did the paperwork for my application about 2 years ago so I don't remember what the design parameters were anymore but basically I wanted to improve on what they (imported equipment) had, both sonically and power wise. Do you have a technical explanation why 10e is better and what effect a 22n cap will have instead of the 100n?
Regards
Chris DC
- Status
- Not open for further replies.