Thank you Jan for reminding us that whenever something like this is implemented or wish to implement the number 1 tool to use is a scope.
Preferably a 100 Mhz. or better for just this reason.
An oscillation induced via a low ESR cap is easily found with a scope. I got my nice HP dual trace 100 Mhz. 1740A on the 'bay for 125 bucks with shipping. It is a must for any audio DIY project.
Now, having said that I have used a Pana FC 100uf cap along with a 1000 pf ceramic on the PS pins of my CDP op amp successfully with no oscillation. Lucky perhaps. This was after I installed a socket with an LM4562 to replace the original op amp.
After socketing the LM it sounded much better and then I added the caps. Result was a small but noticeable improvement in the smoothness I guess you could call it.
I used to be of the no ceramic cap in my stuff fanatic also but I noticed a decrease in the smoothness of the music. Reluctantly I put them back and I have'nt looked back since. Like anything if properly implemented it will work fine, if not the whole mess goes down the drain.
I live in an RFI prone area so I guess this is what I am snuffing out with the proper use of ceramics.
Regards, Marcus
Preferably a 100 Mhz. or better for just this reason.
An oscillation induced via a low ESR cap is easily found with a scope. I got my nice HP dual trace 100 Mhz. 1740A on the 'bay for 125 bucks with shipping. It is a must for any audio DIY project.
Now, having said that I have used a Pana FC 100uf cap along with a 1000 pf ceramic on the PS pins of my CDP op amp successfully with no oscillation. Lucky perhaps. This was after I installed a socket with an LM4562 to replace the original op amp.
After socketing the LM it sounded much better and then I added the caps. Result was a small but noticeable improvement in the smoothness I guess you could call it.
I used to be of the no ceramic cap in my stuff fanatic also but I noticed a decrease in the smoothness of the music. Reluctantly I put them back and I have'nt looked back since. Like anything if properly implemented it will work fine, if not the whole mess goes down the drain.
I live in an RFI prone area so I guess this is what I am snuffing out with the proper use of ceramics.
Regards, Marcus
Guys,
Low ESR/ESL cap at the supply pins of opamp won't make it oscillate, only better ripple rejection. Try to use smallest size SMD closest to the supply pin.
The reason for regulator to oscillate with low ESR cap is because the cap is in the feedback loop of the regulator, which affect the overall compensation.
Regards.
Low ESR/ESL cap at the supply pins of opamp won't make it oscillate, only better ripple rejection. Try to use smallest size SMD closest to the supply pin.
The reason for regulator to oscillate with low ESR cap is because the cap is in the feedback loop of the regulator, which affect the overall compensation.
Regards.
Ipanema said:
That is true, but doesn't change the fact that with some esr you get real damping and can stop oscillations.
In the case of the supply decoupling, if you use low/zero esr caps, the only thing you do is to shunt the supply ripple, noise etc to other resistive elements in the chain like pcb tracks. What you get is that ripple etc as an input signal because it gets in series with the ground ref of inputs or feedback nodes.
In general, if you want to decouple or damp, you need some way to take out energy from the system. You need some resistive element. Ceramic caps are OK for that, cheap, small and they work. Even if they have some tempco, or voltage coefficient, that doesn't matter because the value generally isn't critical anyway.
In applications where the signal is directly affected by cap value or cap parasitics, ceramics are a bad idea, I agree. In feedback or miller compensation, for instance, you should use a very good cap.
Jan Didden
WOW! What a respons to my latest post! Thanks a lot to you all.
So, now I am going to solder a 1nF ceramic cap directly on the supply-pins of the LM4562 opamp. Would this be an good idea? An SMD cap would be difficult to solder because it has no pins to connect it with the opamp.
I can't wait to hear the Tjoeb playing without any HF-distortion, cause its sounds already very very nice right now. This LM opamp realy outperforms the original OPA2604.
So, now I am going to solder a 1nF ceramic cap directly on the supply-pins of the LM4562 opamp. Would this be an good idea? An SMD cap would be difficult to solder because it has no pins to connect it with the opamp.
I can't wait to hear the Tjoeb playing without any HF-distortion, cause its sounds already very very nice right now. This LM opamp realy outperforms the original OPA2604.
Hi,
Current flows in loops and it 'wants' to take the path of least resistance.
What you are doing by adding a decoupling capacitor between the supply pins and gnd at the opamp is providing an alternative path for the current flowing through the supply and ground loop.
At frequencies where the impedance of the decoupling capacitor is lower than the impedance presented by the opamp the current will prefer to flow through the capacitor.
If you look at the impedance plot of a capacitor it is V shaped, high at low frequencies then decreasing with frequency until at the bottom of the V the capacitor is purely resistive (minimum ESR) then as the frequency increases the capacitor starts to look like an inductor and the impedance increases again. This is why you want to minimise the inductance since it reduces the effectiveness of the capacitor to decouple at high frequencies.
Also as well as decoupling, the capacitor will form a low pass filter on the supply line. Let's assume that the resistance of the supply wire is 0.1R, the filter will have a -3dB point at 16MHz, 160MHz and 1.6GHz for 100nF,10nF and 1nF respectively.
Looking at the impedance plots of a typical X7R 0805, the minimum ESR is given at the following frequencies
100nF 35MHz 0.008R
10nF 100MHz 0.100R
1nF 300MHz 0.600R
So, using 100nF is a good start. If you have higher frequency noise in your system then you can also bypass this with smaller values but be careful when connecting caps in parallel that you don't create resonances.
Cheers,
Jon
Current flows in loops and it 'wants' to take the path of least resistance.
What you are doing by adding a decoupling capacitor between the supply pins and gnd at the opamp is providing an alternative path for the current flowing through the supply and ground loop.
At frequencies where the impedance of the decoupling capacitor is lower than the impedance presented by the opamp the current will prefer to flow through the capacitor.
If you look at the impedance plot of a capacitor it is V shaped, high at low frequencies then decreasing with frequency until at the bottom of the V the capacitor is purely resistive (minimum ESR) then as the frequency increases the capacitor starts to look like an inductor and the impedance increases again. This is why you want to minimise the inductance since it reduces the effectiveness of the capacitor to decouple at high frequencies.
Also as well as decoupling, the capacitor will form a low pass filter on the supply line. Let's assume that the resistance of the supply wire is 0.1R, the filter will have a -3dB point at 16MHz, 160MHz and 1.6GHz for 100nF,10nF and 1nF respectively.
Looking at the impedance plots of a typical X7R 0805, the minimum ESR is given at the following frequencies
100nF 35MHz 0.008R
10nF 100MHz 0.100R
1nF 300MHz 0.600R
So, using 100nF is a good start. If you have higher frequency noise in your system then you can also bypass this with smaller values but be careful when connecting caps in parallel that you don't create resonances.
Cheers,
Jon
Hello all,
Do not know about damping requirement, but I can tell you what has worked wonderfully for me with this chip in a Njoe Tjoeb 4000 and other units.
I take a 1uf stacked film cap (panasonic) and a .01uf ceramic cap and put them accross the top of the chip with leads soldered to the supply pins - with the leads as short as possible. Has sounded darn nice every time so far even with senheiser headphones and quite critical listening by others.
Regards,
Greg
Do not know about damping requirement, but I can tell you what has worked wonderfully for me with this chip in a Njoe Tjoeb 4000 and other units.
I take a 1uf stacked film cap (panasonic) and a .01uf ceramic cap and put them accross the top of the chip with leads soldered to the supply pins - with the leads as short as possible. Has sounded darn nice every time so far even with senheiser headphones and quite critical listening by others.
Regards,
Greg
Thanks for your reply, you both.
I still use this player although i've changed the Lm4562 for a Burr Brown OPA2107, which sounds a bit more polite even without the bypass.
@syklab; i dont know if it is directly coupled to the opamp. The tube stage is on a separate print, but the opamp is still on the main Marantz CD-38 print. It's duty is taking care of a part of the filtering.
Soon after my first post, i bought a used upsampler Njoe Tjoeb 4000 with 3 super Tube-clock modules. Its amazing, the most natural, non digital, sounding CD-player i've heard.
http://www.puresound.be/uploads/RTEmagicC_njoetjoeb3.jpg.jpg
I still use this player although i've changed the Lm4562 for a Burr Brown OPA2107, which sounds a bit more polite even without the bypass.
@syklab; i dont know if it is directly coupled to the opamp. The tube stage is on a separate print, but the opamp is still on the main Marantz CD-38 print. It's duty is taking care of a part of the filtering.
Soon after my first post, i bought a used upsampler Njoe Tjoeb 4000 with 3 super Tube-clock modules. Its amazing, the most natural, non digital, sounding CD-player i've heard.
http://www.puresound.be/uploads/RTEmagicC_njoetjoeb3.jpg.jpg
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.