Elevating Signal/Power ground from Chassis Ground

[Deke609]

@PB - I have some very slight hum coming from my Beepre. I want to see if elevating the signal and power ground from the chassis using your reverse parallel diode mod will get rid of it. (I have flux bands wired to chassis ground on the 2 PTs and the 4 PS chokes I added - and am speculating that they might be dirtying up the signal a bit via chassis ground).

As best as I can tell, signal and power share the same buss ground wiring and connect to chassis ground at the inside terminal strips near the 4 pin sockets (on the terminal that is bolted to the chassis - one for each channel).  On the same ground terminals there are 475K resistors connecting the +ve signal output wiring to ground.  Their presence makes elevating the signal/power grounding a bit trickier than I'd prefer.

Question: Can I relocate the 475K resistors to the rca output jacks - tying center pin (+ve) to ground tab (-ve)? Besides ease of installation, is there any other reason for locating them on the terminal strips by the 4 pin sockets?

[Edit: I should have stated that it's not regulator dropout. I know what that sounds like. This is much fainter and gets a little bit louder as I increase mains voltage with the variac. And I did the usual things to figure out that it is in fact the Beepre - swapped/changed tubes, changed cables - no change until I disconnected the power amp from the BeePre (tested with both Kaiju and SII-45)).

many thanks in advance, Derek

 

[Deke609]

I managed to tie in the resistors to elevated ground without relocating them - but it didn't help. Now it's just in the right channel, but louder. The hum gets worse if I touch the top of the right channel 300B, and shows up in the left channel if I touch the top of the left tube.

I figure it's a ground loop. But connecting the BeePre chassis to the Kaiju chassis doesn't change anything.

I'll keep searching and maybe redo some wiring.

cheers, Derek

 

[Paul Birkeland]

Yeah, this is part of the adventure when you stray off the beaten path, you'll have to do a fair amount of experimentation to provide your own debugging. 

You can trim down the level pots on the Kaiju to reduce any residual noise coming out of the BeePre, then just turn the BeePre up a bit.

 

[Deke609]

Yeah, this is part of the adventure when you stray off the beaten path.

Yup. I've been waiting for something like this to happen. I've gone over the wiring half a dozen times and can't see any problems, but reflowed a lot of joints just in case. At this point any more disassembly-then-reassembly is probably only risking more harm.  You can only fiddle with wires so much before they break or loosen. 

All voltages check out, and the amp sounds fantastic - can't hear the hum except in silent passages.  I measured the DC filament "pos out" voltages a few moments ago. Right (offending) side: 9.84 VDC out (0.355 VAC). Left (ok) side: 9.9 VDC (0.345 VAC). And that's at 116.5 VAC mains - they'll be a bit higher @ 120 VAC by way of the variac I've been using to bump up my mains. And I'm within 4% of Jac's filament votage target for the EML 300Bs, so that checks out too. On the off chance: could the 1/3 V of AC on the filaments cause noticeable hum?

I have one Hail Mary left: the hum sounds a lot like the motor hum of one of my aquarium filters that I recently got going again. I've read that electric motors can put junk on the lines. But I'm not holding my breath: I'm powering the amps through a DIY common/differential mode mains filter that is supposed to filter out most of that kind of noise (and it's not the mains filter itself - I tried swapping it out).

If its not the aquarium thingy, I suspect there's some hidden shard of aluminum somewhere that's making a connection that shouldn't be made. Drilling holes in the chassis for the chokes and a couple of new terminal strips produced a lot of aluminum shavings - and some of them went flying off the drill bit. I did a pretty thorough job of shaking them loose and hunting for them, but clearly not thorough enough: I found 5+ tiny aluminum shavings this evening. Three had found there way into the bearing grease of the BeeQuiet attenuators! Which I thought was a most improbable place for them to end up. If they can find their way in there, they could be anywhere. And they are almost impossible to spot b/c they look identical to the chassis.  So I think the best way forward (if it's not the aquarium filter) is to rebuild it.  Maybe even with new filament reg boards if those can be ordered - I did lot of playing around with them trying to get the amp dialed in for the EML 300Bs, and I don't think the traces can take any more rework and are probably on the verge of failing already.

On the upside: I can make the chassis bigger to fit all the stuff I've added.  ;D  At the very least that will make debugging a bit easier. As it stands, I have to remove the giant 22 mF caps every time I want to investigate something.

cheers, Derek

 

[Deke609]

On the off chance: could the 1/3 V of AC on the filaments cause noticeable hum?

I may have the answer to my own question, but hoping @PB for confirmation: the remaining AC on the filament heater wiring doesn't matter b/c it's whisked away to ground via the high capacitance cathode resistor bypass caps. Is this right? Is this the reason for the outsize 10 mF stock capacitance?

I've decided to rebuild the BeePre on a big chassis (16" x 19") b/c I want to incorporate dedicated filament transformers for the EML 300Bs that require 5V 1.3A.  What I need to figure out is how best to do this.

@PB or @PJ: can the LM1085 board handle more than 13.5V input (e.g., 14 or 15V?) and a current draw of 1.3A? I can upsize the wattage/voltage ratings of the resistors and caps and add a larger heatsink to the LM1085 if that will make this possible.

cheers and many thanks in advance, Derek

 

[Paul Birkeland]

The regulator board will have better thermal performance if you cut a rectangular hole in the chassis plate for each board and let the heatsink poke up through that hole.  The heatsink will cool much better when it's not enclosed under the chassis and the ~12V DC that's present there isn't concerning from a safety perspective.

 

[Deke609]

Many thanks PB. So just to confirm: the LM1085 regulator circuit can handle a bit more supply voltage and 1.3A current, provided that adequately rated parts are used and additional cooling is provided via heatsink cutouts? 

cheers and thanks, Derek

 

[Paul Birkeland]

The heatsink is 4.5C/W.  You are dropping about 3.7V across the 1085 at 1.3A, which is 4.8W, we'll call it 5.  This gives 22.5C of temp rise on the heatink and 78C or so working backwards to the junction of the regulator.  The junction shouldn't run any hotter than 125C, and we shoot for less than 100C.  This is too much heat IMO for this heatsink, even poking up through the chassis.  If you had some forced air cooling, this wouldn't be a problem.  There's also a slightly longer version of this heatsink, but I couldn't immediately find any numbers for it. 

Less raw voltage is the easy solution to cool things off.  The 1085 will operate under these conditions, but you'll probably burn the board and things won't run for very long before they fail.

 

[Deke609]

Ah. Many many thanks for this PB. This a great help. I will explore my options - battery powered fans being one of them.

cheers and thanks, Derek

 

[Paul Birkeland]

The raw ~12V DC rails that you're making to heat the BeePre will certainly power a computer fan or two.

 

[Deke609]

Many thanks PB. Tying in fans to the heater power supply would make things easier, but I'm concerned about them injecting noise into the line (I don;t know if they in fact do, but that's what I've read). I'm at pains to minimize all potential sources of noise - hence the battery powered approach.

cheers and thanks, Derek

 

[Paul Birkeland]

Well, you have a running BeePre right now, clip lead one in and see what happens.

 

[Deke609]

Well, you have a running BeePre right now, clip lead one in and see what happens.

Actually ... I don't. I've already stripped it in preparation for the rebuild. :-[ New 16" x 20" chassis from Landfall Systems already ordered.  I hope to have the rebuilt BeePre up and running in about 2-3 weeks and will try the fans then. I also plan on comparing a different dc filament regulator (Rod Coleman's) just to see - but that may take some rejigging of things (different filament transformer, etc.). Again, just to see.  I think I've reached the point where I enjoy the tinkering/experimenting more than listening -- with listening being a momentary reward after tinkering.

cheers and thanks, Derek

 

[Deke609]

It just occurred to me that the new BeePre chassis presents me with another option: mount the LM1085 case directly to sidewall of the chassis - a 16" x 5.25" piece of aluminum. With non-electrically-conductive pad or paste.  I could even add a 1/4" thick, 1" wide copper bar (of any length from say 4" to 15") between the LM1085 and the sidewall for better heat spread. 

Anyone have any thoughts on this?  Seem like a good/bad idea?

cheers and many thanks, Derek

 

[Paul Birkeland]

The heatsinks we provide are far better performers than a chassis plate.

 

[Deke609]

Thanks PB. Hmm. I was hoping the fact that the chassis wall would be exposed to external (cooler) air would offer some advantage. And I thought for sure that adding a copper bar would offer superior performance. I'll need to read up on heatsinks and heat dissipation formulas.

cheers, Derek

 

[Paul Birkeland]

My general rule of thumb is that a chassis plate is about as good as a 40C/W heatsink.  If I have a component that is barely too hot running in free air, then tabbing it down to the chassis can resolve that kind of issue.  If I have a component that is overheating on a decent heatsink, then connecting it to the chassis will make things much worse. 

PJ provided me some information about this some time ago, though I don't have it handy.  If you have a multimeter with a thermocouple and a DC bench supply, you can get TO-220 resistors and do the measurements yourself of the case temperature in free air vs. the case temperature mounted on an 1/8" piece of AL.

 

[Deke609]

The heatsink is 4.5C/W.  You are dropping about 3.7V across the 1085 at 1.3A, which is 4.8W, we'll call it 5.  This gives 22.5C of temp rise on the heatink and 78C or so working backwards to the junction of the regulator.  The junction shouldn't run any hotter than 125C, and we shoot for less than 100C.

How about the Ohmite  FA-T220-64E @ 3C/W for a TO-220 package? Datasheet attached.  That makes 15C rise instead of 22.5.  I don't know how you calculated backwards to the junction, but assuming a fixed ratio of approx 3.5X, that makes 52.5C rise instead of 78.

Haven't measured whether it will fit the existing board - but I could McGyver my own out of protoboard or try to learn Eagle PCB layout if necessary.

 

[Paul Birkeland]

You're going to have to buy it and see if it will even fit on the board.  Forced air cooling makes a lot more sense IMO, though fans may make their own issues.

 

[Deke609]

Yeah, that looks to be both the more effective and simpler solution. So I've ordered two 140mm 12V fans that the manufacturer states produces less than 16 dB of noise at full power - which seems pretty quiet.  They can also be powered @ 7V or 5V, presumably for even quieter operation. I've decided to power them via a separate DC power adapter. They're 25mm thick. But with 5.25" high chassis and some 1.5" feet, I should be good.

cheers and thanks, Derek