Effect of supply bypass positioning.

[Arx]

I'm building a bass guitar amp based on the Basic BPA-200 schematic in National's app note 1192. (The one without the servos)

I'll be making a few small changes. I'll be using 6 3875s instead of 4 3886s, and obviously I'll be building a preamp.

I'm wondering about the effects of having the supply bypass local to the chip vs. at the power supply.
Has anyone done any actual measurements to see how much capacitance is useful?

I'll be wiring it point to point, and I want it to be extremely tough physically. In my last point to point layout I attached the capacitors directly to the chip, supported by their own leads and with the 14ga magnet wire I used for power and grounding. It worked quite nicely, but I want this amp to put up with the occasional hard drop on the pavement without requiring any repairs, so a different approach is needed.

I'm planning to put the bulk of the capacitance away from the chips (probably 30cm or so at most)

How much capacitance has to stay at the pins of the chip? can I do without entirely, assuming the wires are of reasonable guage, and twisted to minimize inductance?

If there's not enough capacitance at the chip, what will happen?

-Nick

 

[Nuuk]

I have used as little as 100 uF on the pins of the chip (when using larger caps 'upstream').

I usually attach the caps to the pins of the chip using their leadouts, and then use hot-melt glue to secure them in position.

In your case, I would make the amp modules and then find a way to insert them in the case such that they are partially floating!

 

[martin clark]

If there's not enough capacitance at the chip, what will happen?

Oscillation or instability under load. It will not necesarily be audible.

If your main PSU caps really are 30cm away I'd seriously consider a few tens of uF at the pins as well as the HF bypass that NUUK suggests.

 

[Ponz]

the rule of bypass capacitor is as closer as possible, and using mutiple value of capacitor together to get wide range of freq. response.
for example: the bypass capacitor can use 100uF + 0.1uF PP capacitor or 10uF + 0.1uF and attach capacitor's pin directly to the power amp ic.

 

[Nuuk]

I'd seriously consider a few tens of uF at the pins as well as the HF bypass that NUUK suggests.

Am I missing something here Martin? A few tens is not as much as 100?

 

[Gordy]

Oscillation or instability under load. It will not necesarily be audible.

It may not be directly audible as oscillation per se, but may cause loss of detail and smearing of the sound. Also, it may cause the chip to over-heat and enter thermal shut-down.

Capacitors as close to the chip as possible = a good thing.

 

[Arx]

I have used as little as 100 uF on the pins of the chip (when using larger caps 'upstream').

I usually attach the caps to the pins of the chip using their leadouts, and then use hot-melt glue to secure them in position.

In your case, I would make the amp modules and then find a way to insert them in the case such that they are partially floating!

Is 100uF the minimum you've gotten away with, or the minimum you've tried?

I'm also hoping to keep electrolytics away from the heatsink. I think I'll probably just try and minimize the distance to the bigger caps, and see if I can get away with a 4.7uF film cap at the leads. (Of course, that's starting to get heavy again... I might need to go with the ugly hotglue thing)

-Nick

 

[Nuuk]

Is 100uF the minimum you've gotten away with, or the minimum you've tried?

I know I tried 100uF but I have a feeling I also used 47 uF with a regulated supply. But look, the difference in size/weight between these smaller caps is really insignificant!

I also build with the caps kept off the heatsink!

 

[Minion]

I usually put 1000uF per rail on the amp boards and 100nF Film caps right at the power pins of the chip to ground......

So far my Guitar amps runs like Champs.....


Cheers

 

[BWRX]

How much you put close to the chip depends on a few things - what frequencies will the chip amplify, how much output power do you expect the chip to produce on a continual basis, and of course size/layout contraints. To maximize space you can use 470uF 50V caps that are about as tall as the chips, otherwise I would side with Minion and try to use up to 1000uF 50V per rail. Of course, place any smaller film/ceramic decoupling caps as close to the chip pins as possible.

If you're looking for ruggedness you really should use a PCB.

 

[Arx]

How much you put close to the chip depends on a few things - what frequencies will the chip amplify, how much output power do you expect the chip to produce on a continual basis, and of course size/layout contraints. To maximize space you can use 470uF 50V caps that are about as tall as the chips, otherwise I would side with Minion and try to use up to 1000uF 50V per rail. Of course, place any smaller film/ceramic decoupling caps as close to the chip pins as possible.

If you're looking for ruggedness you really should use a PCB.

Size isn't really a problem. I'm more concerned with heat and inertia. Obviously with the chip right beside the heatsink, the capacitors aren't going to be much further away.

Maybe I'll give the PC board a thought, though my heatsink is about 30cm square.

To distribute the 6 chips across the heastink I'll likely want the chips mounted at the edges of the board, on the same plane, so I'll have to bend a lot of pins...

-Nick