[Folsom]

I've been using an Audiosector 3875 kit for a lot of years now. (actually built numerous amplifiers with Peter's kits)

My right channel made some weird, quiet, RF like noise, that slowly got closer and closer together. Now nothing comes out of the right channel... I guess they just only last for so long!

But now I can't decide to I just replace it? Or is it time to move on?

I have Pass F1 boards sitting in my closet. But they are an expensive build. (not even sure if I can find a pre-made transformer)

Was also looking at 41hz units as well. Sigh...

[*Scotty*]

DoS, the chip is only $6.95 plus shipping from Mouser
http://www.mouser.com/ProductDetail/Texas-Instruments/LM3875TF-NOPB/?qs=sGAEpiMZZMvu8NZDyZ4K0dxenBCCX%252bVU
You might try putting a 400pF cap in series with a 400ohm resistor on the input from hot to ground to give the amplifier some protection from RFI. As it sits there is nothing to stop RFI from getting into the chip. This needs to be kludged right on the board and not at the RCA jacks.
Scotty

[Folsom]

I have 300pf caps on pins 7 & 8 I think it is, already. I get RF anyways but don't know why. (I've done everything there is possible to avoid it, grounding, caps, etc, can't figure it out)

I know the chip is cheap. But sometimes it is good to move forward too. I might like something else more.

[poseidonsvoice]

I have 300pf caps on pins 7 & 8 I think it is, already. I get RF anyways but don't know why. (I've done everything there is possible to avoid it, grounding, caps, etc, can't figure it out)

I know the chip is cheap. But sometimes it is good to move forward too. I might like something else more.

Your other option, which is more expensive but can provide galvanic isolation, is to use an input transformer like I did:









Best,

Anand.

PS. I built Peter's design, years ago, a great amplifier indeed! The Class D Audio SDS 254 module beat it however. Then the Ncore 400 arrived, and that beat the SDS 254. On a rating of 1-10, the LM3875 was about a 7, the SDS 254 was a 9, and the Ncore was a 12 

[Speedskater]

Well the Audiosector 3875 is one of those minimized parts count defective designs.  They remove the protection components and cross their fingers. 

[Folsom]

The nCore looks really great but They are a little too expensive for my current budget. (maybe a little expensive for the parts on them too)

I'm not sure where the RF gets into my Gainclone. I'm not so sure it is the RCA's.

[jtwrace]

The nCore looks really great but They are a little too expensive for my current budget. (maybe a little expensive for the parts on them too)

I'm not sure where the RF gets into my Gainclone. I'm not so sure it is the RCA's.

Buy a preowned UcD then. 

[*Scotty*]

The resistor in series on the input with the cap to ground can lower the filter pole of the RF filter by factor of ten increasing its effectiveness greatly. Depending on where you live the filter pole 6dB/oct filter you now have may not be enough.
I wouldn't say there anything terribly wrong with the design, the application notes on the entire Overture series neglect to mention the need for a RFI filter on the input of the chip and Patek wasn't aware that there could be a problem. Even if you don't think you can hear it, the RFI will show up as fur on the waveform and the RF filter can reduce or remove the fur depending on where you are located.
Scotty

[Folsom]

Buy a preowned UcD then.

I don't follow?

ScottyThe resistor doubles it huh? That is something I surely would do.

There are multiple RF sources in town (radio stations). At my old abode I could hear the rock station a few blocks away in the bathroom fan.

[Folsom]

Come to think of it, RF is probably why it went bad...

You know, ferrite beads, star grounding, nothing helps that I can tell for cutting the RF. The capacitors helped some, but not nearly enough. I think I'll upgrade to 400pf, with resistors on both. Perhaps I'll replace both chips at the same time. (going to be a lot of un-soldering going on)

[*Scotty*]

If Anand's build is the "beauty" for "gainclones" which is probably the case, then mine must be the "beast"and plain in the bargain.
$55 chassis from ParMetal, 625watt Avel Lindberg toroid transformer and a total of 20,000ufd per rail of Jensen 4 pole electrolytic
storage caps. I used the AudioSector LM 4780 board.


The zobel on the output in my amp is 4.7 ohms and .1uf. This differs from the schematic specified resistor for this position which is 2.7ohms.
Scotty

[poseidonsvoice]

The nCore looks really great but They are a little too expensive for my current budget. (maybe a little expensive for the parts on them too)

I'm not sure where the RF gets into my Gainclone. I'm not so sure it is the RCA's.

Okay, let me put it into more perspective, as I've listened to the Hypex UCD and neglected to include it in my summary! Silly me!

LM3875 -> 7/10
SDS 254 -> 9/10
UCD 180HG HxR ->10+/10
NCore 400 -> 12/10.
NCore Bridged 400 -> ??
NCore 1200 -> ???

Buy a preowned UcD then. 

Both UCD and NCore are made by Hypex. But you can buy preowned UCD's pretty easily as it is an older design, as jtwrace points out. Or you can build one. It's on my list for a 7 channel amplifier.

If Anand's build is the "beauty" for "gainclones" then mine must be the "beast"and plain in the bargain.

You are too kind sir

Best,
Anand.

[Folsom]

Scotty that is... Hey do you know if I can increase the feedback resistance to get less output impedance? Would that work? I've been thinking about trying to tame my Horn Shoppe Horns a bit more. The F1 offers that, but I thought GC might be able to with changes to the feedback.

How do I search for UCD modules? Used? Not seeing much out there.

[*Scotty*]

DoS, if you download the PDF doc for the LM 3875 and check the pin out on the chip you will see the positive and negative rail voltage inputs, the inverting and non-inverting signal inputs which make possible bridged operation and the speaker load output pin.
There is no way to alter the feedback used in the amplifier that I can see. You can do some things to increase its stability like adding the RFI filter I mentioned as well as increasing the resistor value in the zobel network used on the output but other than doing this, the chip is kind of a black box device. You put your signal in one end and it comes out bigger on the other end, you really can't monkey with it between input and output.
Scotty

[wakibaki]

If the implementation looks anything like the datasheet then there will normally be 2 resistors which control the gain via feedback, Rf (usually 20k) and Ri (usually 1k). This gives a gain of 20/1 = 20. The more feedback (the smaller Rf, the lower the gain) the lower the output impedance. Increasing the resistance (Rf) increases the output impedance, because the feedback is reduced. Some designs also use Rf2 and Cf to roll off the high frequency gain still more. Most of these Overture chips don't like the gain reduced below 10, and this is the case for this one too, so the scope for reducing it isn't great. If you reduce Rf, you're unlikely to see a beneficial effect on your speakers, the principal difference will be the loss of volume.

RF can get into an amplifier in a lot of different ways. On the speaker leads is common. Filters are often installed. One of the main assets in the fight against RF is the box (if metal). Pass several turns of any wires that enter the box through a toroid core positioned as close as possible to the box. Type 73 material is good up to ~30MHz, other materials may be required. You may be able to determine the frequency of the interference by identifying the program, and hence it's source (station). Here's a page with good information about which kind of cores to use:- http://www.cwsbytemark.com/CatalogSheets/Ferrite_datasheet_oct06/FR_MATL.pdf.

I'd want to fix the amplifier to keep as a spare or to use elsewhere even if I decided to build another one.

[Folsom]

I had that backwards. If I increase the resistance, I raise the impedance and therefor lower the dampening factor (what I want). But I don't want to increase RF, that is a bad idea.

I'll stick to the 400pf and resistor. Although it probably just isn't the correct amplifier for me.

Personally I don't like the sound of ferrite beads and the like on my music, anywhere near the RCA's.

Have to keep my eye out for Hypex stuff. Part of me likes tube amplifiers, but getting any kind of power on them is a nightmare.

[wakibaki]

Personally I don't like the sound of ferrite beads and the like on my music, anywhere near the RCA's.

Well, I certainly don't have ferrite beads on my guitar strings, although I do confess to having had a few coloured beads in the past, but only on the loose bits above the tuning machines. Anyway, I suggest a toroid, not a bead.

Which is worse, the interference you can hear or the ???????????, whatever you think the ferrite might cause. You'll never know unless you try it.

I wouldn't recommend something that was likely to have a negative effect on the sound.

Ferrites such as I'm talking about are lossy at radio frequencies but transparent at audio, and I'm talking about the speaker and power cables. They carry large power levels, so any effect the ferrites might have is likely to be invisible, due to the large ratio between the signal (or the power) and any possible distortion caused by the ferrite. They will add some inductance to the cable/speaker assembly, but a speaker is a pretty big inductor, often with a large core of sintered metal, not that different from ferrite. Of course the speaker is at the wrong end of the wire to prevent it acting as an antenna.

On the other hand, the tiny levels of RF, which ferrites can effectively block, can easily cause large audible disturbances when they get into an amplifier, and are also thought to be responsible for degrading amplifier sound in a non-obvious fashion, i.e. you may not hear somebody going 'Breaker, breaker', but it might be causing a general lack of fidelity (due to intermodulation) of the sound that you will only notice when it's gone (if even then). Because intermodulation can take place between the RF and the audio, the effect of the RF itself can be unfelt, until the music starts.

If you want to decrease the damping factor, you can do this with a series resistor between the amp and the speaker.

w

[Quiet Earth]

If it were me I would move on.

If I had been using that amp for many years, and always during that time trying to figure out different ways to battle RF, i think i would just throw in the towel. Especially now that a device has blown. The thing is probably oscillating. There's probably something fundamentally wrong with the design or the amp was not made with the correct parts to begin with. Either way, unless you get that ironed out,,,,  like with a schematic of a known problem-free version,,,  your just shooting in the dark.

One other thought.  Do you have a zobel on your single driver speakers? Something to consider for when you get your new amp.

Best of luck man. I know the feeling for not wanting to let it go.

[Folsom]

Well I'm going to try a 41hz Amp10.

I need to get some information on it to know if there are any signal path components (for me to replace with high end stuff), working on that. I got a kit with enclosure for great price from someone on forum. Then I got a transformer from another forum member (700va, over-over kill but big enough I can put in an F1J if I want to build up those boards later).

[Folsom]

I may have found what caused it.. cold solder on the speaker out post. Remarkable that it could have been, but it was... I'm pretty good at avoiding them but I guess it can happen. Well I'm sure the stinking RF in the area doesn't help.