Hi Mike,
For bypass capacitors for your power supplies, I'd go with ceramic capacitors. As much as I hate them for audio, they are one of the best for this purpose.
Where you ground your bypass caps is just as important as using them or not. The ground these return to is a "dirty" ground. All the supply noise currents will be returned here, so you don't want that in any signal ground references.
-Chris
I ran a high end repair shop for 16 years. I sold it about 10 years ago. It was 6,000 sq. ft with a warehouse and about 10 employees. Now I work by appointment and do Cyrus and Creek warranty service for Canada.
For bypass capacitors for your power supplies, I'd go with ceramic capacitors. As much as I hate them for audio, they are one of the best for this purpose.
Where you ground your bypass caps is just as important as using them or not. The ground these return to is a "dirty" ground. All the supply noise currents will be returned here, so you don't want that in any signal ground references.
-Chris
Is that 2 - 1 to the Ceramics then!? 😉
Or is an injury time equalizer on the way?
Chris, your point about the return path has me 'scratching head' a bit. I mentioned earlier that there are two ground paths associated with the LM3886 in my amp. The decoupling caps ground to the 0v on the transformer secondary which has a spur off the the ground on the main smoothing cap. The ground pin on the chip goes all the way around the rest of the circuit via each area. It eventually ends up at the RCA inputs then the regulator grounds, and finally ends up back at the main smoother.
I would guess that the 'dirty' route would be the longer of the two. If this is the case then its a good thing as this enables me to solder these caps directly to the chips pins (+ve to gnd, and -ve to gnd) as suggested earlier.
I have so far seen that both my Cambridge CDP and Amp use small 47nF Polyester caps on the opamps and as bypass on the smoothing caps around the regs. This is lower voltage application that the power chips but does it suggest that Ceramics arent the only option for local decouping bypass duties?
Whats the comon type of cap you have seen in the Cyrus and Creek gear you have serviced?
Mike.
Or is an injury time equalizer on the way?
Chris, your point about the return path has me 'scratching head' a bit. I mentioned earlier that there are two ground paths associated with the LM3886 in my amp. The decoupling caps ground to the 0v on the transformer secondary which has a spur off the the ground on the main smoothing cap. The ground pin on the chip goes all the way around the rest of the circuit via each area. It eventually ends up at the RCA inputs then the regulator grounds, and finally ends up back at the main smoother.
I would guess that the 'dirty' route would be the longer of the two. If this is the case then its a good thing as this enables me to solder these caps directly to the chips pins (+ve to gnd, and -ve to gnd) as suggested earlier.
I have so far seen that both my Cambridge CDP and Amp use small 47nF Polyester caps on the opamps and as bypass on the smoothing caps around the regs. This is lower voltage application that the power chips but does it suggest that Ceramics arent the only option for local decouping bypass duties?
Whats the comon type of cap you have seen in the Cyrus and Creek gear you have serviced?
Mike.
the 'dirty' ground is basicly the ground thats connected to all the power supply related components (i.e the smoothing caps).
The signal ground will have to connect to it somewhere with a small/narrow connection path.
So it sounds to me that the gnd connected to the RCA and the regulator is the clean signal ground
The signal ground will have to connect to it somewhere with a small/narrow connection path.
So it sounds to me that the gnd connected to the RCA and the regulator is the clean signal ground
The ground on the left where the decoupling caps are is the one that goes direct to the trafo secondary and negative pin on the main smoother. The one on the right goes from the gnd pin of the LM3886 via the everything else. If Ted is right then I cant solder the caps directly to the chips pins as I intended.
The thing is, all the ground traces end up at the same place anyway.
Hi Mike,
Install those capacitors across your existing capacitors.
Your clean ground is what runs to the RCA connectors. Your amplifier will amplify the difference of signal between these two points, so you certainly don't want any noise there.
Most electronic equipment uses ceramic capacitors for power bypass. They have lower inductance than almost any other type of capacitor, which is what you want. Surface mount capacitors have the lowest lead inductance (guess why! 🙂 ), but are easily damaged through normal soldering techniques due to thermal shock. So use radial leaded devices and keep the leads short.
You will notice that the common ground from those two capacitors go only to the star common point. You know what that says to me?
Dirty ground! 😀
Imagine that all traces are low value resistors with a little inductance thrown in. Add some capacitance connecting them to adjacent traces. If you see things that way, the concept of grounding will be easier to understand. This also explains why we want to get rid of higher frequency noise at the source. I'll answer that one for you. The higher frequencies transmit in air and between adjacent traces much easier than low frequency noise. That's in general mind you. That is one simplified reason why three terminal regulators do so much better at isolating hum and ripple than they do with bursts of oscillation (from the diodes as an example).
This goes to the concept of modifying equipment. There is no one best type of capacitor (for example). You use the best type of part that is suited to the job at hand. So in this case, ceramic capacitors are the best suited to the job of bypassing high frequency noise to ground. More properly, bypassing keeps the effective power supply impedance low at all frequencies along with some other, larger capacitors.
-Chris
Install those capacitors across your existing capacitors.
Your clean ground is what runs to the RCA connectors. Your amplifier will amplify the difference of signal between these two points, so you certainly don't want any noise there.
Most electronic equipment uses ceramic capacitors for power bypass. They have lower inductance than almost any other type of capacitor, which is what you want. Surface mount capacitors have the lowest lead inductance (guess why! 🙂 ), but are easily damaged through normal soldering techniques due to thermal shock. So use radial leaded devices and keep the leads short.
You will notice that the common ground from those two capacitors go only to the star common point. You know what that says to me?
Dirty ground! 😀
Which begs the question. When is a ground, not a ground?
Imagine that all traces are low value resistors with a little inductance thrown in. Add some capacitance connecting them to adjacent traces. If you see things that way, the concept of grounding will be easier to understand. This also explains why we want to get rid of higher frequency noise at the source. I'll answer that one for you. The higher frequencies transmit in air and between adjacent traces much easier than low frequency noise. That's in general mind you. That is one simplified reason why three terminal regulators do so much better at isolating hum and ripple than they do with bursts of oscillation (from the diodes as an example).
This goes to the concept of modifying equipment. There is no one best type of capacitor (for example). You use the best type of part that is suited to the job at hand. So in this case, ceramic capacitors are the best suited to the job of bypassing high frequency noise to ground. More properly, bypassing keeps the effective power supply impedance low at all frequencies along with some other, larger capacitors.
-Chris
Hi,
you can mount the decoupling caps direct on the +ve and -ve pins.
The other two legs connect together and float below the PCB. Now connect the SHORTEST possible wire link from that floating common connection to the NEAREST Power Ground.
I read that the current coming from the chip ground pin is not clean.
The Author wrote that it had pulses and noise and he recommended it was not returned to the Signal Ground.
you can mount the decoupling caps direct on the +ve and -ve pins.
The other two legs connect together and float below the PCB. Now connect the SHORTEST possible wire link from that floating common connection to the NEAREST Power Ground.
I read that the current coming from the chip ground pin is not clean.
The Author wrote that it had pulses and noise and he recommended it was not returned to the Signal Ground.
I'd love to redesign this PCB with relay input switching and seperate star grounds for Power and signal (Naim style). Id paralell a pair of LM3886 per channel for more power. I'd also use local regulation for each section and keep signal routing as short as possible using a pot extension..........Same parts just a better PCB layout would probably make a big difference!
Mike.
Mike.
Yeh, I think it would.
The 4870 (iirc) IC is already 2 x LM3886 in parallel (~100W RMS @ 4Ohms). These can be bridged and/or paralleled again for even more power!
Simon
The 4870 (iirc) IC is already 2 x LM3886 in parallel (~100W RMS @ 4Ohms). These can be bridged and/or paralleled again for even more power!
Simon
Cool! Or should I say hot!!!😎
Do they have the same (or close) pinout as the LM3886? (?Guess what Im thinking?)
Joking. 😉
Maybe sometime I'll look into designing my own PCB's.
Do they have the same (or close) pinout as the LM3886? (?Guess what Im thinking?)
Joking. 😉
Maybe sometime I'll look into designing my own PCB's.
there's only one way to find out.mikesnowdon said:
Are you prepared to cut that trace and hardwire a connection to the Power Ground?
mikesnowdon said:Cool! Or should I say hot!!!😎
Do they have the same (or close) pinout as the LM3886? (?Guess what Im thinking?)
Joking. 😉
Maybe sometime I'll look into designing my own PCB's.
I dunno, Google for LM4870 datasheet 😉
There's not a lot of point, IMHO, in upping the chip's capabilities, retaining using such a tiny power transformer.
I'd not bother with a PCB, it's just extra work and extra places for signal quality to get lost. Hard-wire two LM3886s or 4870s, use a nice, big transformer (e.g. 500VA toroid from Airlink - £30), a chunky bridge and good PSU caps (lots to choose from) and you'll have a very capable amp.
Now marry this with the Pedja Rogic discrete buffer or similar, powered by S Power regulators, and you will forget the Cambridge ever existed and be glad you sold it on for parts funds... There's also then scope for some fun DIY case-work, relay input switching, elaborate volume control, that sort of thing..
That's just my opinion.
Simon
Apparently there is a very expensive commercial 'high-end' amp using 6 x LM3886 per channel! (Borat) Niiiice, I like! 😎 😀
Details:
Details:
An externally hosted image should be here but it was not working when we last tested it.
Mark Levinson? I think there's something to be said for current ability / headroom and low impedance 🙂
Your transmission lines probably enjoy powerful designs, which may help to keep that big bass in check.
Your transmission lines probably enjoy powerful designs, which may help to keep that big bass in check.
Yep. I didnt know Levinson were using the LM chip too. Got a link or any pics of the insides. I love to see the guts grrrrrrrrrr! 😀
Not without proper study and planning prior.
So far wherever I have seen additional bypass to the decoupling caps - soldered direct - to the LM3886, guess which caps were used? Wima 0.01uf. Im not saying that its wrong to use ceramics but if people in general go for the wima then I'll do the same.
Mike.
- Status
- Not open for further replies.