Bypass caps, worth it ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Gooday all,
What is the last word in bypass caps on the main caps in the power supply. Is it worth the effort ? If so, what would the value be on two paralleled 6800uf caps per rail and what make and value ? New to the diy setup The cover is off the amp to replace output relay so I can add bypass caps at the same time.I have read the threads on the topic and still do not know if left from right, such a lot of different opinions.

Thanks for your input.

Jan


Regards
 
You probably have already something on the amp PCB and before rectifiers.
Most manufacturers add a 0.1uF on each rail (to ground) on the amplifier board. It prevents the HF noise from the rectifiers to propagate in the rest of the system (using the wires inductance).
Some manufacturers (Accuphase) add also 0.1uF on the main capacitors.
 
Last edited:
bypass caps

Thanks for the reply's you guy's, I am learning from you people, I want to get it right the first time. I like it a lot more to relax and listen to to music that to struggle to get it right .I am reading a lot and are amazed by the wealth of knowledge that you have. You are more clever than a tree full of owls ! :D DF 96, just trying to piece everything together.

Regards
Jan
 
Upuesto Paul,
Thanks for the tips,I will incorporate it. One learns from listening, everybody learns everyday something new .Hehe, you're doing just fine, so weird to hear that phrase ! My dad used to use it to say goodbye to friends after a visit. Never heard it again in 22 years.Are you originally from South Africa ? or does it come from the dutch ? (Als van die beste vir jou en jou familie )

Regards
Jan
 
I had 0.1uF on my main caps, then 1R + 0.1uF also in the PSU, and an umbilical to the amp. When connected to my PC, there was crazy hum, which entirely went away when I removed everything but the bulk capacitors. This is with a centre-tap transformer setup.

On the amp boards there is 1,000uF || 10uF || 0.1uF.
 
This is a complicated matter.

Some people report good results when using smaller value bypass caps, others not so good or bad results. IMHO the reason is all to do with LC resonances between the inductance of the wires or tracks connecting the caps and the caps themselves and any resistance in the circuit which could act to damp those resonances. Since all these factors will be dependent on the particular amp layout, there's no definitive answer as to which is better I'm afraid.

My experience is that if you use good quality electrolytics close to the active devices with a bit of damping resistance you don't really need smaller value bypasses. In fact they can make things worse by exciting new resonances.
 
Last edited:
I'd agree KatieandDad, it's worth a try. The point I was trying to make is that the results may vary from one person to another, so don't expect it will work for you just because it did for someone else, even if the cap values are the same.

The other thing to consider is don't go for the boutique caps or ultra-low ESR caps. Bog standard ones often do a better job because their higher ESR adds a bit of damping to any resonances around. Believe me, you're not hearing the silk fibres or bamboo splinters or whatever version of snake oil the caps are filled with!
 
Last edited:
I think that, even better than bypass caps, are two banks of caps connected by low R (e.g. 0.33R) 5W wirewound resistors like what is shown in the attachment (it's the PS from NP's F5 Turbo)

This arrangement forms an RC lowpass filter that can be designed to have 6dB to 10dB of attenuation at 100/120Hz (the ripple frequency) and then keeps on attenuating as frequency increases. This can really reduce noise on the rails.

Seemed topical to mention this here. Any downsides apart from using 2x the PS caps?

-Charlie
 

Attachments

  • F5_Turbo power supply.PNG
    F5_Turbo power supply.PNG
    74.2 KB · Views: 176
Charlie

There are no downsides other than a small loss in voltage due to the resistors, which in this case is fairly negligible even at full current. The RC filter plays a much more important role than just reducing the actual 100/120 Hz ripple level. It cuts out a lot of the higher frequency ripple harmonics that would otherwise get through to the audio circuits and cause problems. If you look at the ripple waveform before the resistors it will be a sawtooth shape. After the filter it will be much more sinusoidal ('ish).

All the caps after the bridge contribute to reducing ripple, it's not that you have to double them up. You're just splitting the number you would normally use into two groups separated by the paralleled resistors.

Inductors aside, this is probably the best approach to heavy duty power supplies. I might add another set of paralleled resistors between the bridge and 1st bank of caps to reduce the charging current and for additional ripple reduction. Also, some recommend making the 1st bank of caps half the capacitance of the second bank, again to reducing charging currents. Add a snubber after the 2nd bank of caps a la CarlosFM and bob's your uncle.
 
Last edited:
Charlie

There are no downsides other than a small loss in voltage due to the resistors, which in this case is fairly negligible even at full current. The RC filter plays a much more important role than just reducing the actual 100/120 Hz ripple level. It cuts out a lot of the higher frequency ripple harmonics that would otherwise get through to the audio circuits and cause problems. If you look at the ripple waveform before the resistors it will be a sawtooth shape. After the filter it will be much more sinusoidal ('ish).

All the caps after the bridge contribute to reducing ripple, it's not that you have to double them up. You're just splitting the number you would normally use into two groups separated by the paralleled resistors.

Inductors aside, this is probably the best approach to heavy duty power supplies. I might add another set of paralleled resistors between the bridge and 1st bank of caps to reduce the charging current and for additional ripple reduction. Also, some recommend making the 1st bank of caps half the capacitance of the second bank, again to reducing charging currents. Add a snubber after the 2nd bank of caps a la CarlosFM and bob's your uncle.

Hi Gopher,

Thanks for the follow up. I knew about the RC filter action at high(er) frequencies but thanks for mentioning it. Yet another advantage of this design.

Also, thanks for your thoughts on charging currents. I hadn't thought of that aspect of the design. I think that the only downside of reducing the total capacitance of the first bank is that you will need to increase the series R to get the same level of filtering at the ripple frequency. If you are primarily interested in high frequency filtering of noise and diode ringing then this is not necessary. On the other hand if you are primarily looking for ripple reduction there is point at which the resistance of the series R will start to increase the total impedance of the rail. From what I have seen, this won't happen unless R is more than 1 or 2 ohms.

-Charlie
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.