Grounding, hum, etc.

[cpaul]

I've had a problem with my Q producing hum in most systems I put it in.  I know there are impedance issues with some amps that have somewhat low input impedance (10k to 50k), and gain issues with most, but I get hum with all of them.  Had some success pairing it to a 6V6 tube amp at one point a year or two ago.  I posted about this issue quite a few years ago, then shelved it.  Recently, in an effort to revive Q (probably mostly as a headphone amp) I decided to read up on this forum.  I read about a solution connecting the Q ground buss (on the RCAs) to the chassis of the amp. 

Since the amp RCAs are eventually grounded on the amp (pretty sure) I figured that couldn't make a difference (b/c the interconnect connects the RCA grounds anyway).  But then I thought it through and figured it couldn't hurt.  So I did it.  Ground should have no voltage on it (unless my filament batteries are backward, but I'm pretty sure I fixed that), nor does the chassis of the amp.  It' a LM3886 gainclone and has a 3-prong outlet with a star grounded chassis.  Should be OK.

I hooked up a wire to one of the RCA grounds and clipped it to the chassis of the amp.  NO NOISE.  Great.  Then smoke.  Lots of smoke.  The ground wire I'd added melted off the insulation.  Hmmm.  LOTS of current there. 

So I measured voltages between the RCA buss and the chassis of the amp with Q on. I'm getting about 1.5v there (Q positive relative to the amp).  Then I turned on the amp and measured.  I got 33v or so with Q being positive relative to the amp.  Note the Q was on for both measurements.

What?  How could that change?  Q should be at the same potential which makes it seem to me the amp chassis goes negative when the amp is turned on.  But I've measured and the amp chassis is at mains ground.  I'm thoroughly confused.  Any ideas?  I wouldn't be surprised if 33v was the current B+ voltage on my partially used batteries.  But why wouldn't it measure that way with the amp off?

Thanks a ton in advance for any ideas.
Carl

[mcandmar]

On your gain clone measure from the chassis (i.e. mains ground), to the RCA ground (audio ground) and see if there is a difference.

Purely speculating here, but if your gainclone has a split rail + and - power supply with approx 33v on each side,  and the - side tied was tied to ground you would get those symptoms.   i.e. the entire amps ground reference is out by 33 volts.

Since the quickie isn't referenced to mains earth it wont care, but whatever source your feeding into the quickie wont be happy if its tied to ground as you will have a 33volt differential between the two.

[Doc B.]

Try turning the D cells around.

[cpaul]

Thanks guys.  mcandmar, the gainclone does, of course, have a split supply and I don't recall what the voltage is at the moment.  But it may have been a 24v PS iron (48vct, rectified into a +24/-24 supply) which would be around 33-34v (+/-) after rectification and smoothing.  Yours is a great suggestion and I'll have a look and report back (though the gainclone itself has NO noise on it's own and when used with another tube linestage).  Doc B, I'll also check the filament batteries by swapping them.  I built mine right around the time the filament battery instruction issue came to light, so I recall checking that several times when I built it.  But I'll re-check since I still had 1.5v when the gainclone amp was off.

I REALLY appreciate the input!

[cpaul]

Well, here's an update.  Mcandmar, you were spot on.  Just measured the amp on it's own and here's what I found.  The + supply is at 66v relative to the chassis (!), the "ground" for the PS is at 33v, and the - supply is at zero.  Again, that's measured relative to the chassis.   Meaning the chassis is floating at -33v.  Sheesh.  What I had forgotten is that I built the gainclone using an old Magnavox chassis for a DVD/HT amp and just re-used the old TWO prong plug.  The chassis is NOT grounded (I have a wooden front plate on it and the chassis was hidden in the rack it was on, so I guess I didn't worry about it at the time).

I still don't get why the measurements changed with Q on vs off, but I'm pretty convinced that's the problem.

Anyway, that's the crux of the hum/buzz problem AND the melted ground wire to Q.  Hope I didn't do any damage to the Q.  I had pulled the amp out of circulation except for this test, so it's back to the bench to fix that amp.  I measured a bit and don't see any neg 1.5v on Q so I'm guessing there isn't a problem with that.  I'll further test that and may re-wire anyway as I'm thinking of some upgrades.  Once I get the gainclone fixed I'll try again with the ground connection.  But based on that -33v in the chassis going through that wire, I'm pretty sure I'll get less hum/buzz.

[Paul Joppa]

Can't pass up mentioning the **battery powered** Quicksand! Uses a zero-prong outlet.  :^)

It appears the chassis is somewhere connected to the negative supply. Is there a reason it does not go to the signal ground (power supply midpoint)?

[cpaul]

Can't pass up mentioning the **battery powered** Quicksand! Uses a zero-prong outlet.  :^)

Hahaha!  Love it!  And I've considered it.  But my budget is micro and I really need more power for my Vandersteen speakers.

The PS output has 3 terminals (it's a +/- supply after all) and all 3 terminals are connected to the correct points on the amp boards (I'll double check that).  The input RCAs are isolated from the chassis as are the speaker terminals, so I can't for the life of me figure out how the chassis is connected to the negative rail.  But I'll work on it.

On another note, I plugged the Q into my main amp last night and had a listen.  It's microphonic as usual but the buzz is gone.  There's a faint deep hum that is only noticeable when I'm near Q, and it tends to fade but it's probably always there a little.  More importantly, I did a short listening session (maybe 45 minutes) and MAN was I pleased by what I was hearing.  Nice to get back to Q.

Carl

[Paul Birkeland]

I smoked a Gainclone in a similar fashion, though it wasn't with the Quickie.  High feedback solid state gear can be persnickety, and implementing some bandwidth limiting filters at the input of the gainclone, as well as current limiting devices may be quite helpful for you.

-PB

[cpaul]

Thanks, Paul.  I'll consider your remarks.

[cpaul]

OK, for anyone following this, here's what I found.  The cheap power supply PCB I had for the gainclone (PS and two signal boards cost me $7 shipped...from China so it's probably a rip-off not to mention slave labor.. live and learn) has 4 mounting holes that are plated with relatively large metal pads for mounting screws.  It all looked fine, with the boards being surprisingly nice (OK, if you accept that a gainclone isn't really a gainclone if it's on a PCB!).  I had mounted the PS board with some star lock washers to keep the mounts tight.

Well, one or two of the lock washers were wider than the mounting hole pads by a tiny bit, so they touched the green PCB coating.  I'm assuming they scraped off the coating and made contact with some tracings which happened to be ... the negative rail.  In short, the non-insulated mounting screws were creating continuity between the chassis and the negative rail.  Never mind that the negative rail shouldn't be anywhere so close to the chassis mounts in the first place - such a situation shouldn't be possible IMHO. 

I replaced the metal lock washers with some nylon washers and the problem disappeared.  No continuity, no voltage on the chassis.  I then replaced the power cord with a 3-conductor cord and grounded the chassis.  Connected the previously floating center point on the PS to the chassis star ground, and fired it up.  No problems.  And Quickie plays quite nicely with this amp now.  Some slight hum, which I suspect may have to do with the low input impedance of the gainclone, but will experiment.

in short, though, (no pun intended) the key problem has been fixed.  Thanks for help folks!

[Grainger49]

I'll bet that the input resistor for the GC can be adjusted to what is needed.

[cpaul]

Yes, indeed, Grainger.  That's my next step.  Figure out which is the input resistor and play with it.  I installed one resistor with standoffs that let me change it easily.  I assume that's the one but need to pull out the schematic or the LM3886 data sheet to figure it out.