Raspi 2 and HiFiberry DAC+ coupled to a LM3886T Gain Clone Amp

[jls001]

I have kept the mains ground open for now - . Wanted to put a bridge rectifier and close the loop. Referring the the doc

You can use a 35A bridge rectifier later to ground your equipment, but for now till you can arrange the bridge connect the ground from IEC socket to both chassis (wire to be terminated in a lug and lug to be screwed down to BARE chassis by scraping off the paint or anodization, and use washers and two nuts to interlock securely), and transformer chassis.

The hifisonix document is very comprehensive and I admit I don't follow all the points mentioned, especially "hum breaking resistors" which I'm hearing for the first time.

You are operating very hazardously right now. Do be careful not to cause electrical shock to yourself as there is no return path right now.

 

[omishra]

You are correct. When i tried creating the overall circuit diagram, i could see where i am creating a loop. Not sure how to avoid it though.

At the input, when i connect a stereo cable, it invariably shorts the two input grounds and needless to say that should be causing the loop issue. The million dollar question is...how to resolve it?:sad:

I was looking at your diagram. It seems okay. But how you implemented it, is the question.
Assumptions:
1. All your RCA and banana plugs are isolated from chassis.
2. DC ground is isolated from chassis.

Now important part starts.
1. Don't share power +/- and ground wires among amp boards. Take separate wires from PS for each amplifier.
2. The ground, specially from mid of capacitor bank.
3. Input ground lands on amplifier boards only. No shortcuts.

If you do this then i will be interested in hi res picture of wiring. So far keep chassis grounded to earth but DC ground lifted. We will discuss ground isolator later.

 

[omishra]

I have Separate Power Supplies for both Channels (Jean Hiraga 30 Watts) and i have faced this hum issue. When i run single channel there is no hum, but as soon as i connect the 2nd channel RCA the humming starts.

I tried various methods to eliminate the hum, I did brought down the hum within acceptable limits, but didn't succeeded to completely eliminate it. I also tried by disconnecting the ground from 1 channel but was in vein.

I am using Cap Multiplier (2 Nos) each for single channel.

Any suggestions Mishraji?

Hi Sadik, sorry for late reply. Above steps applies to you plus you need two ground isolators made from bridge rectifier and one 10 ohm, 10A NTC.
Connect star ground of each PS to chassis through ground isolator. Don't mix both star grounds at chassis without isolator. With isolator both grounds are separated by ~4V probable difference. So keep both DC grounds isolated and separated always. I and many are using this scheme successfully in may class A amplifiers.

Some pictures are in my old post. When I bolt this isolator then take care bolt grounds only appropriate connection.

Take clue from figure 4 on this page..

 

[sunilmani]

I was looking at your diagram. It seems okay. But how you implemented it, is the question.
Assumptions:
1. All your RCA and banana plugs are isolated from chassis.
2. DC ground is isolated from chassis.

Now important part starts.
1. Don't share power +/- and ground wires among amp boards. Take separate wires from PS for each amplifier.
2. The ground, specially from mid of capacitor bank.
3. Input ground lands on amplifier boards only. No shortcuts.

If you do this then i will be interested in hi res picture of wiring. So far keep chassis grounded to earth but DC ground lifted. We will discuss ground isolator later.

Assumptions:
1. All your RCA and banana plugs are isolated from chassis.
sunilmani==> Yes
2. DC ground is isolated from chassis.
sunilmani==> yes

Now important part starts.
1. Don't share power +/- and ground wires among amp boards. Take separate wires from PS for each amplifier.
sunilmani==> Yes, they are separate wires.
2. The ground, specially from mid of capacitor bank.
sunilmani==>capacitor bank is PCB mounted and has a connector. The ground is from there.
3. Input ground lands on amplifier boards only. No shortcuts.
sunilmani==>input ground lands on the pcb itself. no shortcuts.

Chasis has not been grounded to earth. Doing it now. Was busy over the weekend to implement suggestions. Will try out the chasis earthing.

 

[sunilmani]

Last night i found some time for myself and i was hurriedly trying out the grounding. I had the stereo cable connected to RCA with the stereo jack left open. Had the speakers connected. The earthing from the supply was connected to the torroidal mount screw using washers....There was strangely a lot of hum..more than the usual one..i could hear from the speaker.

I was yet to procure the bridge rectifier, so went without it. Then i...connected DC ground to chasis :-( . There was some sparking when i did that and everything went silent. I quickly removed the DC ground connection to chasis. I connected the stereo pin to my laptop and played some music, things worked normally. There was nothing out of place. Not sure what happened there but strangely all the humming was gone; as though there was some capacitance accumulated that caused the hum and earthing DC ground discharged it. Hope i have not damaged the chips...

 

[sadik]

I connected the stereo pin to my laptop and played some music, things worked normally.

Was your Laptop Charger connected to AC supply (Charging Mode) while playing the music?

 

[sadik]

Hi Sadik, sorry for late reply. Above steps applies to you plus you need two ground isolators made from bridge rectifier and one 10 ohm, 10A NTC.
Connect star ground of each PS to chassis through ground isolator. Don't mix both star grounds at chassis without isolator. With isolator both grounds are separated by ~4V probable difference. So keep both DC grounds isolated and separated always. I and many are using this scheme successfully in may class A amplifiers.

Some pictures are in my old post. When I bolt this isolator then take care bolt grounds only appropriate connection.

Take clue from figure 4 on this page..

Yes i have done it the same way.. but still there is a slight hiss sound (high frequency). Its not actually a humm (low frequency) sound.

 

[sunilmani]

Was your Laptop Charger connected to AC supply (Charging Mode) while playing the music?

No. The laptop was not on charging at that time.

 

[omishra]

Yes i have done it the same way.. but still there is a slight hiss sound (high frequency). Its not actually a humm (low frequency) sound.

Hiss is different thing.
It comes from 3 points.
1.Wires picking from surrounding or input output crosstalk. This you may have avoided by routing and quality wires. Test this with input grounded at each - pre or power amplifier to see where it gets in.
2. High gain stages either in pre or power. Adjustment of volume control and levels can give relief.
3. Inherent design issues or layout which affects feedback implementation and produce oscillation as white noise. We call it hiss.

 

[sunilmani]

Finally some free time. The sparking was found to be caused due to the IC body touching the heat sink. I have put in 1 mm silicon sheet in between and its working good. Somehow not satisfied with the way i have mounted the IC's to heat sink using the silicon sheets.

Have powered up the RASPI as well which is running ELEC. The combination is working great.
The completed amplifier is looking as follows
Front View
Top View

Do we have any documents/methods specified as to how to use a silicon sheet for connecting to heat sink?

 

[jls001]

I have put in 1 mm silicon sheet in between and its working good. Somehow not satisfied with the way i have mounted the IC's to heat sink using the silicon sheets.

Hope you've used heat sink paste with silicon layer? The paste helps with more efficient heat transfer from chip/power transistor to heat sink. I usually apply the paste to both sides of the silicone sheet.

In general, the metallic body of the LM3886 must not be in galvanic contact with the heat sink (though heat sink may be anodised, it will have electrical contact with cabinet through mounting screws). So the screw that holds the chip to to heat sink must also be electrically isolated from the metallic body of the chip through a plastic ring.

 

[sunilmani]

Yes. I have applied a fine layer of heat sink paste to the chip before mounting it on to the silicon layer; i have used a nylon washer as well to isolate between the screw and chip; but the problem i am seeing is due to tightening the screws i believe. I am using a M3 screw and on tightening the silicon sheet just bends over. Guess i am tightening it more than what is required. Are there any guidelines on how much to tighten these screws?

 

[tcpip]

Since you're asking highly specific questions, I'll chip in. You're already facing the problems some of us have faced.

I don't use the hole in large chips for fixing them. "Large chips" includes amp chips like LM3886/3875 etc, and also large OPS transistors used in power amps.

For all of them, I use a half-inch wide strip of aluminium. I cut it to a length maybe 20mm longer than the chip width. I drill two holes on the two ends, and I use two bolts to clamp the strip on top of the chip.

This has many advantages. Firstly, I don't need to hunt for shoulder washers, which are hard to find. Second, I get to apply pressure across the _face_ of the chip, not just at the top. This allows me to press the full surface of the chip down on the heatsink, and makes excellent thermal contact. Third, I can use a single aluminium strip to clamp down three or four devices, e.g. a power amp which had two OPS devices per rail, requiring me to clamp down four large TO3P transistors to the heatsink. Fourth, I can standardise on M3 bolts and taps to do the drilling, tapping and final fixing. One size fits all.

Hope this helps. Wish I could show you a photo.

 

[tcpip]

Clamps are also very useful. These things have a spring action to press the device down onto the heatsink.

 

[sunilmani]

Since you're asking highly specific questions, I'll chip in. You're already facing the problems some of us have faced.

I don't use the hole in large chips for fixing them. "Large chips" includes amp chips like LM3886/3875 etc, and also large OPS transistors used in power amps.

For all of them, I use a half-inch wide strip of aluminium. I cut it to a length maybe 20mm longer than the chip width. I drill two holes on the two ends, and I use two bolts to clamp the strip on top of the chip.

This has many advantages. Firstly, I don't need to hunt for shoulder washers, which are hard to find. Second, I get to apply pressure across the _face_ of the chip, not just at the top. This allows me to press the full surface of the chip down on the heatsink, and makes excellent thermal contact. Third, I can use a single aluminium strip to clamp down three or four devices, e.g. a power amp which had two OPS devices per rail, requiring me to clamp down four large TO3P transistors to the heatsink. Fourth, I can standardise on M3 bolts and taps to do the drilling, tapping and final fixing. One size fits all.

Hope this helps. Wish I could show you a photo.

One more clear advantage is that the holes on the chip need not perfectly match on heatsink. I was running into these issues as well.

 

[tcpip]

One more clear advantage is that the holes on the chip need not perfectly match on heatsink. I was running into these issues as well.

Ah, yes, that's a problem. I solve it by never soldering the chip to the PCB early. I first finish soldering everything else on the PCB, then I fix the chip to the heatsink, then thread its legs into the holes in the PCB, and finally I solder the chip's legs to the PCB. This ensures that there is no strain on those legs or the solder joints.

 

[sunilmani]

Ah, yes, that's a problem. I solve it by never soldering the chip to the PCB early. I first finish soldering everything else on the PCB, then I fix the chip to the heatsink, then thread its legs into the holes in the PCB, and finally I solder the chip's legs to the PCB. This ensures that there is no strain on those legs or the solder joints.

I do not have the option of doing that since the legs are to be soldered from below the PCB and there is no space to do the same.

 

[Nitin Arora]

I am envisaging a problem here. Since the media center is within the box, the power off may not always be controlled and also it may not be preceded with turning off the media center. The result maybe a corrupt database and that would leave Kodi unusable for next time...any thoughts? There are suggestions on super capacitors in some forum...there should be a better way to do this...

In almost 2 years of being a member, I never thought I could be of any help here. :lol:

Please let me know if you are still trying to resolve this. I was in a similar situation while building a media center of a Rpi2. I found it really irritating to have to plug the power out and then back in to have to restart the pi. Plus I guess I had some spare time to kill. So I made a simple board using an ATMega328 (essentially a standalone Arduino), since that was what I had lying around to be able to power on and power off the pi using a push button.

I essentially used two power supplies, one to power the ATMega board and another to power the pi which is controlled by a relay board controlled by the ATMega. If the pi is off, the circuit essentially closes the relay booting up the pi. If the pi is on, the circuit pulls down (or up cant remember, not that it matters) a GPIO on the pi which has a python script running causing the system to halt. This halt is detected by the AT using the UART pin on the pi to trigger an interrupt which is used to poll the pin about 10 times (to ensure that this is indeed a shutdown) and the open the relay. The advantage of this approach is that one can simply push the button to shut the system down or use the software interface (OpenELEC in my case) to power down and the system still works.

I have the eagle pcb lay out so if you are willing to home etch the board then i can send it over along with the code. You will need some way to program the ATMega along with a few additional components and some tweaks to the code since you are using a HiFiberry.

 

[sunilmani]

In almost 2 years of being a member, I never thought I could be of any help here. :lol:

Please let me know if you are still trying to resolve this. I was in a similar situation while building a media center of a Rpi2. I found it really irritating to have to plug the power out and then back in to have to restart the pi. Plus I guess I had some spare time to kill. So I made a simple board using an ATMega328 (essentially a standalone Arduino), since that was what I had lying around to be able to power on and power off the pi using a push button.

I essentially used two power supplies, one to power the ATMega board and another to power the pi which is controlled by a relay board controlled by the ATMega. If the pi is off, the circuit essentially closes the relay booting up the pi. If the pi is on, the circuit pulls down (or up cant remember, not that it matters) a GPIO on the pi which has a python script running causing the system to halt. This halt is detected by the AT using the UART pin on the pi to trigger an interrupt which is used to poll the pin about 10 times (to ensure that this is indeed a shutdown) and the open the relay. The advantage of this approach is that one can simply push the button to shut the system down or use the software interface (OpenELEC in my case) to power down and the system still works.

I have the eagle pcb lay out so if you are willing to home etch the board then i can send it over along with the code. You will need some way to program the ATMega along with a few additional components and some tweaks to the code since you are using a HiFiberry.

The catch here is that there is no separate push button and power supply. The mains power off needs to be detected and shutdown initiated; also the raspi needs to be powered up for whatever time it takes to shut down.
I am still grappling with this. Have couple of solutions, but none which is simple and cost effective.

 

[sunilmani]

Since you're asking highly specific questions, I'll chip in. You're already facing the problems some of us have faced.

I don't use the hole in large chips for fixing them. "Large chips" includes amp chips like LM3886/3875 etc, and also large OPS transistors used in power amps.

For all of them, I use a half-inch wide strip of aluminium. I cut it to a length maybe 20mm longer than the chip width. I drill two holes on the two ends, and I use two bolts to clamp the strip on top of the chip.

This has many advantages. Firstly, I don't need to hunt for shoulder washers, which are hard to find. Second, I get to apply pressure across the _face_ of the chip, not just at the top. This allows me to press the full surface of the chip down on the heatsink, and makes excellent thermal contact. Third, I can use a single aluminium strip to clamp down three or four devices, e.g. a power amp which had two OPS devices per rail, requiring me to clamp down four large TO3P transistors to the heatsink. Fourth, I can standardise on M3 bolts and taps to do the drilling, tapping and final fixing. One size fits all.

Hope this helps. Wish I could show you a photo.

I have taken a closeup of the PCB mount. It will give a fair idea of what the situation is. Here is the link.