Millett "Starving Student" hybrid amp
Aug 18, 2010 at 6:33 PM Post #5,431 of 7,277
I agree with TomB. The shoulder washer completely isolates the tab from the case. I think it's time you show us a couple high-res pics from your amp's innards :)
 
 
Aug 18, 2010 at 8:16 PM Post #5,432 of 7,277
Well, there is definitely another issue in my amp aside from my mosfets. even with the plastic screws added, something causing a huge spark as I turn the amp on. The LEDs still light up, so I don't think it's the mosfets that are shorting (initially when the mosfets were shorted, the LEDs didn't turn on). I searched for "spark" in the thread but didn't find anything enlightening.
 
EDIT 2: I disconnected the mosfets again, let them hang, and turned on the amp - no spark. I guess it really is an issue with my mosfet mounting. I suspect it has something to do with the hole I drilled for the mounting of the mosfet. I sized it for slightly larger than a 4-40 screw, but the mosfet/Bergquist pad hole is larger than that. The holes in my top panel are not the cleanest, and I had a number of burrs with the smaller holes. There might be a small bump at the edge of the hole that is coming up and contacting the mosfet flange. Guess i'll have to run out to Radioshack and pick up some a new thermal pad or two.
 
I thought I had been meticulous enough in planning the build that I wouldn't have any miswirings. I'm guessing if there is an issue, it is either with one of the panel components or something is shorting as I cram the thing inside the case.
 
Here are some pics i've taken:
http://picasaweb.google.com/submergence/SSMHTroubleshooting?authkey=Gv1sRgCPnjg6DLuZLykAE#
I've taken some pics of my general wiring, as well as pre-casing pics of the real guts of the amp. I've added a lot of crap in the panel components, though, so the wiring gets very confusing very quickly. I also used solid core 22awg wire, so things stay put, but on the flip side are difficult to move.
 
As for my Ground - I was originally going to use a large copper plate as ground and solder all my ground connections to that, but after I cased everything up it quickly became evident that I wouldn't have much space for the copper, so I opted to use a star ground and just soldered all my ground wires (ALL of them - LEDs, output jack, pot, power, input, etc) together (black).
 
Red wires are the right channel, white wires are the left channel. blue wires are directly off of power, green wires are either my DAC outputs or LEDs.
 
The reason my amp is such a snake's nest is due to all the extras i've stuffed in there - DAC power, DAC output, DAC LED, two tube LEDs, input selector from RCAs, and a gain switch. I also left all of my leads a good bit longer than necessary should I have to cut something off and redo a connection. I put heatshrink over just about every exposed joint, so it's unlikely that one part is shorting against another part.
 
EDIT: okay, I turned on the amp ensuring that no exposed metal was touching another part of exposed metal. The sparking is coming from the back panel itself - generally from the screw holes, or from the nicks in the anodizing of the back panel. I think this points to a grounding issue, as I think the only part that touches the case metal is the ground (outside) connection of my DC jack. Is that right?
 
Aug 18, 2010 at 11:18 PM Post #5,433 of 7,277
Anyone know where I can get solid thermal pads B&M (read: immediately)? Radioshack's website shows they have some type of thermal pad, but it shows as web-only. I'm hoping a place like Best Buy carries thermal pads for GPU ramsinks, but I doubt it...
 
Aug 19, 2010 at 7:06 AM Post #5,434 of 7,277


Quote:
Anyone know where I can get solid thermal pads B&M (read: immediately)? Radioshack's website shows they have some type of thermal pad, but it shows as web-only. I'm hoping a place like Best Buy carries thermal pads for GPU ramsinks, but I doubt it...


Radio Shack should have them in their small parts bins - at least the TO-220 mounting kits.  You'll probably have to buy one of those.  Even if you go to an electronics supplier like Mouser/DigiKey/et al, you'll have to purchase mica insulators in multi-packs.
 
Best Buy is not going to have anything like this.  Fry's will, if you have one of those nearby.
 
 
Aug 20, 2010 at 10:46 AM Post #5,436 of 7,277
I had to go to two radioshacks to find the mounting kits. The first radioshack I went to (near work) had no idea what they were, and I couldn't find them in-store. Actually, they told me it was an "online only" item. So I got back to work, checked online, and of course - they did show as available at the very store I went to. So I called the store and asked for the catalog number. the person said "hold on, let me go check", then promptly hung up. I called back again, and a different person answered - I asked for the first one, she said "hold on", then put me on hold for five minutes - and promptly hung up. At this point I just gave up.

Fortunately the radioshack near my house DID have them - exactly two in stock. I asked them if they could keep them on hold for me and they said sure, asked me my name, and put them behind the counter.

While at work, I expanded the mounting holes slightly - from 1/8" (.125") to... 1/7? (0.140"), and also made sure to deburr the hole from the inside, as per Tomb's advice. When I got home I used some Arctic Silver Ceramique I had left over from a CPU upgrade (non-conductive). It's messy stuff, and harder to spread than I thought! A thin layer on both sides, and some fine working with tweezers and the shoulder washer... everything is in place. Flip the switch... NO SPARK! whoooo (is it bad I expect a massive spark every time I turn on the amp now?).
 
For posterity, I was going over the thread again last night and I came across this post . I don't know how I didn't catch it the first time around, but it exactly described my problem.
 
Some quick thoughts until I can get to my nicer camera and attach my real volume knob.
     - my tube LED switch doesn't really make much of an impact. I use 4.3kohm and 2.4kohm resistors (each one feeding both LEDs) connected to the switch. However, there isn't much of a brightness difference. Next time I open my amp and do some surgery I think I'll try to get the LEDs at 3mA and 10mA.
     - tomb's "sea-green" LEDs are more "green" than "sea"... his picture is deceptive! 
frown.gif
I want my money back.
     - For better or for worse, the tube heaters are very dim; there's only a tiny hint of a glow at the very top of the tube. I can't see it unless I turn out the lights and turn the tube LEDs off.
     - both heatsinks and tubes get nice and toasty, hot enough that I wouldn't really want to hold either for longer than a few seconds. I'll see if I can't find a temp probe somewhere in the lab to get hard numbers. EDIT: after about an hour of on time, the entire amp gets pretty toasty. I think I'll have to put finding that thermo higher up on my priority list. EDIT2: a surface temp probe tells me the temp is around 40-45C, but I don't believe it - I would say it felt closer to 50 or even 60 C. An 80C oven did not feel that much hotter than my case. I'm a little concerned about running this thing for long periods of time...
     - my diode protection LEDs flash nicely when I turn on and off the amp... and with headphones connected, there's still a rather unfriendly thump, but only when turning the amp off. I guess I'll still be taking precautions when turning the amp on and off.
     - my gain switch does not do a whole lot. I have it switching between 50k and 150k input resistors, along with a 50k amp. I'd say at the top end of volume, there's only about a 20% difference between the two. This is with low-impedance headphones, though, so I don't know how it reacts to, say, Sennheisers or Beyers. I hope to find out soon
beyersmile.png

     - perhaps related to my gain switch, hiss becomes rather noticeable at about the 3'o'clock position for my 150kohms (low gain?) and maybe 1:30 for my 50k (high gain?). This is too bad, because the whole reason for the gain switch was to use more of the pot... oh well.
     - is it just me, or is the pot really cheap? I didn't think i was putting very much stress on it and when I looked last night, the wiper pin had somehow detached itself from the pot. The whole pin! I thought it was a bad solder job at first, but I noticed the heatshrink was still on, and the end of the wire didn't look like a real wire. I put a large blob of solder here to hopefully prevent it from happening again. Just a note for people in the future, especially if you're using solid core wire - be careful with the panel components!
I think that's all for now. Back to the music!
L3000.gif

 
Aug 20, 2010 at 3:09 PM Post #5,437 of 7,277
Quote:
For posterity, I was going over the thread again last night and I came across this post . I don't know how I didn't catch it the first time around, but it exactly described my problem.  

 

Who can blame you? With over 5400 replies in this thread and no sign stopping it's not easy to find these little nuggets.
 
I wonder how Pete Millett feels about the still growing popularity of his design.
 
Aug 20, 2010 at 10:52 PM Post #5,438 of 7,277

 

 

 

 
The rest of the pics (50 decent ones all told) are here:
 
http://picasaweb.google.com/submergence/StarvingStudentMillettHybridSSMH?feat=directlink
 
I haven't done a serious critique of the thing (and I probably won't until I get back to my desktop and proper desk at school), but it sounds quite good. I also noticed that my tube LED pics are the same color as TomBs - very deceptive! This must be the LED's version of bait-and-switch. They look okay for now, but I'd really love to swap them out for red ones, and change RLEDs and gain resistors at the same time. But for now - god knows I've spent enough time on this project.
 
From start to finish, a twinkle in my eye to pushing out Rob Thomas, was about three months total - from reading this thread, to lots and lots of planning, then about a week for construction and casework (which took up most of the time, really - from drawing out exactly where holes would go, measuring inside and outside dimensions of each panel component, lots of pixel nudging in Photoshop until everything fit just perfectly and looked good at the same time, and learning how to use the laser and figuring out how to make the text look like it wasn't spray-painted on). Speaking of the laser, the engraving on the front, back, and bottom panels was done with a VersaLaser engraver we have at work. My boss mentioned it, and I figured if I have access to cool tools, I may as well utilize them. I personally think this finishing touch is what really makes this amp my baby - not the hours and hours of hard and painstaking work, the many schematic interpretations and placement maps I drew, having to carry this thing gingerly in my backpack to and from work every day for two weeks, the hours I spent making sure the holes I drilled with the drill press were EXACTLY where they needed to be, the high I got from the solder fumes, and the stupid MOSFET grounding issue I had. But all of these things is what makes me really proud to be a fath-... er, a DIYer.
 
I'm still a little concerned about the heat this thing generates, though - I'll have to hit it with an IR gun when I get back to work to find out what the real surface temps are. Until then... I'll be testing to see if this amp can play music 24/7, and how long I can listen to it until my ears start to bleed!
L3000.gif

 
Aug 21, 2010 at 6:47 PM Post #5,439 of 7,277
Holy ****!  That looks amazing!  Congrats, sir.
 
Aug 22, 2010 at 12:15 PM Post #5,440 of 7,277
Yes, very nice work!  The laser etching adds a professional touch.
 
About your comments on the "ocean green" LED's - it seems like changing to red is a fairly radical departure from that.
wink.gif
  In any event, the Angstroms are clearly specified (5000):
http://beezar.com/oscommerce2/catalog/product_info.php?cPath=39&products_id=114
There's no attempt at exaggerating the color description.  It comes straight from the supplier.  Plus, anyone can look up that value of Angstroms on a color chart if there's any doubt.  Generally speaking, blue light is considered as wavelengths from 4500 to 5000 Angstroms, whereas green goes from about 5000 to 5500 Angstroms.  There are differences in those bandwidths, depending on which source you reference - some say blue is 4500 to 4900 Angstroms and green from 4900 to 5600.  In either one, though, it's pretty clear that the LED in question is much closer to the blue side of green, hence "ocean green."
smile.gif

 
Aug 22, 2010 at 12:48 PM Post #5,441 of 7,277
Color and color names are in the eye of the beholder. If you do not believe that, walk into any paint store.  :wink:
 
I think it looks nice. Certainly more interesting than Red. Red is so overdone.
 
Aug 22, 2010 at 10:28 PM Post #5,442 of 7,277
Well here it is.
I've named it "The Kryptonite Supercharger".
(It will have green LED's under the base and around the phone jack in the future)
 
No problems at all with this PTP build, it worked flawlessly from the first time I plugged it in.
No static hiss, no hum, no microphonics. It also has pretty good channel matching.
Of course I stand on the shoulders of those who have labored hard through the myriad
of problems encountered and solved, I was able to avoid them. (after reading 360+ pages!)
 
I used a variation of the star ground that I call "the bar ground".
It is basically a 2 inch long piece of 12 gauge copper wire strung between two terminal strips.
The aluminum case is isolated from all ground points except where the terminal strips are bolted.
Towards the end of construction I was running out of ground tie points, so it is a little messy
underneath the chassis with jumper wires running over others. 
 
Lastly, I want to thank all of the principles who contributed their knowledge and time to this project.
Thank you Mr. pmillet for giving this elegant design to the world, thank you Mr. tomb,
thank you Mr. dsavitsk, thank you Mr. n_maher and thank you to the_equalizer who
became quite the expert towards the end of this thread with his tube rolling mods.
 

 
Aug 23, 2010 at 12:05 PM Post #5,443 of 7,277
@Aflac and @livewire: 
 
   Great job on both your builds, they look fantastic !    I really dig the copper pipe and radiator in livewire's amp and I'm impressed with the  hi-tech look of Aflac's panels and lighted tubes. 
 
I hope you enjoy using your amps as much as you evidently enjoyed putting them together. Well done gentlemen !
 
Aug 30, 2010 at 7:45 AM Post #5,444 of 7,277
Hello,
 
I tried my SSMH today, I first only plugged the Power Jack, with no source or headphone. All was working okay: the two tybes were glowing, with the same brightness, i let it like this for around two minutes, then, I plugged an old walkman, and a cheap per of earphones, I couldn't hear any sound, I then hear and saw a spark who was, I think, coming from the power jack, and the tubes started to unglow, and i shut it down. I opened it, tried to debug but all looked fine, i verified some opened wire, and put some heatshrink on it, just in case, I plugged th power jack, i heard and saw again the spark, and immediatly shut it down, there was no glowing tube.
 
Do you know what can be my problem? How can I find out by myself what it is? Some tests to do?

I'm husing a metal Hammond Case, maybe this is part of the problem? I used a blank PCB platform, i don't know if I had to solder a wire between it and the case, so I didn't, indeed, i used a file to remove the black paint, so my RCA jacks are also grounded by the case.
 
Thanks for your help!
 
Antoine
 
Aug 30, 2010 at 11:41 AM Post #5,445 of 7,277
Did you insulate the mosfets from the casing? The metal mosfet tab must not be connected to ground! Check if there's continuety between the tabs and the metal casing. Did you use an insulating washer when mounting the mosfets to the heatsinks?
 
 
Do not use the metal casing as a ground conductor. Use a wire for grounding. Only connect the metal casing to ground at the power input.
 

Users who are viewing this thread

Back
Top