Millett "Starving Student" hybrid amp
Jul 24, 2013 at 2:04 AM Post #6,616 of 7,265

mosshorn

1000+ Head-Fier
Joined
May 23, 2011
Posts
1,233
Likes
140
Quote:
Last Thursday at about 10pm CST, another 12AU7 variant Millett Starving Student was born! I'm thrilled to say that the first power-on went flawlessly, all systems go right out of the gate with nothing to troubleshoot! Dead quiet too, omg. After a weekend of off and on burn in, this thing absolutely sings! I love it! Mates up well with my grubDAC and my slightly modded Grado SR60i's. I did somewhat of a perfboard design to keep the guts from turning into too big of a mess and used a copper board ground plane. It's using some 50's RCA long black plate 12AU7 tubes.


Here's a couple of glamour shots and the obligatory pic of the guts.





 

I broke up the resistors and capacitors into two perfboard sections, essentially one for the tubes and one for the output. They fit in nicely I think!

Cheers!

 
Wow, that yellow is gorgeous! Great work!
 
Jul 26, 2013 at 1:51 PM Post #6,618 of 7,265

Tjj226 Angel

500+ Head-Fier
Joined
Oct 20, 2012
Posts
988
Likes
574
Hey guys, I am looking at building one of these for a friend. Not too long ago I saw a web page for the 12AU7 mod, and now I can not find it. However, as I look through the pages of this post, it seems like you can just build the amp and throw 12AU7s in it and call it done. I am not sure if this is true or not. 
 
Provided that the 12AU7 mod is not a mod at all, does anyone know of a kick arse mod to do to this amp to really make it sound warm, dark, and lush? 
 
Jul 26, 2013 at 3:25 PM Post #6,619 of 7,265

Goobley

100+ Head-Fier
Joined
Aug 10, 2012
Posts
133
Likes
12
It is a slight mod, the bias point for the MOSFET and valve heater has to be moved slightly. Here's the page for the mod with the schematic: http://diyforums.org/SSMH/SSMHvariants.php

This amp sounds good on its own, but if you want a slightly blacker background (also increases the gain slightly) then you could think about adding a constant current source to the tube, this does however add some complexity to the tube stage.
 
Jul 26, 2013 at 11:58 PM Post #6,620 of 7,265

holland

Headphoneus Supremus
Joined
Jun 15, 2007
Posts
2,210
Likes
22
Warm, dark, and lush is the opposite of what a CCS would do.  You sound like you want tube distortion, and lots of it.  Resistors.  The "sound" of the 19J6 at the operating point for the SSMH is warm, but not really dark and lush, even though it's not in the best operating point for the tube.  The 12AU7 mod, I have no idea, but I've heard plenty of 12AU7s and I wouldn't necessarily say it's as lush a a 6DJ8, generally speaking.  There may be a tube that exhibits that, however.
 
Depending on what headphones you're using, you may get the sound you want with a Morgan Jones amp from Headwize, if it's even still alive as a site.
 
Jul 29, 2013 at 11:29 PM Post #6,621 of 7,265

anachronsimic

New Head-Fier
Joined
Sep 25, 2012
Posts
15
Likes
10
Quote:
I apologize for not knowing where to find it specifically, but there is another variant that used metal tubes.  Maybe someone else will point it out.
wink.gif

 
Quote:
tomb: Thx for summarizing! [FYI: My main ref to this project has always been: http://www.pmillett.com/starving.htm ] So, based on your  succinct summary ...this thread seems ridiculously long...i.e.,  if your two external links (Millet's DIY site pages) are the ONLY major topological changes/ mods to the orig. design... then I'm guessing much of the discussions in THIS ~440-page  thread have been ...  ... about what ... MSSH cosmetics :wink: ??
 
IAC ... I'm perhaps most curious about the addition of CCS. Not sure whether KimLaroux's recent mention of it is new to this thread and, in general,  the MSSH design. How does CCS improve sound over orig.? 

 
 
I know that this is probably a moot point at this point but: http://www.head-fi.org/t/319231/millett-starving-student-hybrid-amp/2820#post_5584504
 
The reason I've held on to that link is that I'm actually planning on making that version at some point but it's sort of a pipe dream at the moment.  I'm expecting to actually make it some point this year though, as I'm gonna start my EE education soon.  
 
Anyways, continue.
Edit: Aw man I'm already beat to it. My bad, sorry.
 
Jul 30, 2013 at 6:34 PM Post #6,622 of 7,265

ProTofik

Head-Fier
Joined
Jan 10, 2013
Posts
95
Likes
14
Quote:
 
Your output caps, C3 and C5, are too small.  I used 1000uF, but 470uF is a minimum.  KimLaroux's formula is almost correct.  You need to take into account that the high pass filter is C3 (or C5) interacting with the parallel resistance of 2K and the output headphones.
 
Anyway, the corner frequency, which the equation 1/(2*pi*RC), is about the -3dB point, and it bends about it at 6dB per octave.

 
I replaced these capacitors with 1000uF, 25V ones. Here is the result.
 

 
Thanks. Are there any other recommended modifications that I can make to my student? This is how it looks right now:
 

 
 
And one more thing. I would like to make a linear 48V power supply (I use switching one right now). I know that my amp at start need about 1.5A and it drops after few seconds to 380mA. Can you recommend me what transformer and parts I should use? I know how to make simple LM78xx based power supplies but I never done anything above 15V.
 
 
 
Jul 31, 2013 at 10:30 PM Post #6,624 of 7,265

plumcakk

New Head-Fier
Joined
Jun 21, 2009
Posts
34
Likes
12
Here is my first build! Chose the 17EW8 because it was cheap. Made an absolute mess of the wire management so there's not many pics of the internals (it's not pretty). That'll teach me for not using a ground plane/using #18 in a 4.5" x 7" x 2.5" box!
 
Relatively trouble-free start up with no hiccups, aside from the fact that apparently I wired the pot backwards. So now it turns counterclockwise.
 

 

 

 

 

 

 
 
 
I am curious though, I seem to be hitting my personal volume limit on my DT880/600s with the pot turned maybe 10 degrees, and that's with 50kohms input resistance(the second switch on the top of the box is meant to engage the input resistors). If I want to be able to utilize more of the pot, that means I should try a larger resistance on the input, right?
 
Aug 1, 2013 at 1:58 PM Post #6,625 of 7,265

holland

Headphoneus Supremus
Joined
Jun 15, 2007
Posts
2,210
Likes
22
Quote:
Chose the 17EW8 because it was cheap.
 
I am curious though, I seem to be hitting my personal volume limit on my DT880/600s with the pot turned maybe 10 degrees, and that's with 50kohms input resistance(the second switch on the top of the box is meant to engage the input resistors). If I want to be able to utilize more of the pot, that means I should try a larger resistance on the input, right?

 
Bad choice.  The 17EW8 has a mu of 50.  The 19J6 is about 39, IIRC.  The 12AU7 is around 17-19.
 
Larger resistance will reduce signal, and with noise in the amp getting amplified, your signal to noise ratio decreases.  It'll work, but I think it's a crappy solution.  If you look at Pete Millett's original design with the 19J6, he had some feedback to reduce the gain.  You can add some more, but what you started with is a poor choice.
 
How did you select the operating point of the 17EW8?
 
Aug 1, 2013 at 6:55 PM Post #6,626 of 7,265

plumcakk

New Head-Fier
Joined
Jun 21, 2009
Posts
34
Likes
12
Quote:
 
Bad choice.  The 17EW8 has a mu of 50.  The 19J6 is about 39, IIRC.  The 12AU7 is around 17-19.
 
Larger resistance will reduce signal, and with noise in the amp getting amplified, your signal to noise ratio decreases.  It'll work, but I think it's a crappy solution.  If you look at Pete Millett's original design with the 19J6, he had some feedback to reduce the gain.  You can add some more, but what you started with is a poor choice.
 
How did you select the operating point of the 17EW8?

 


Yeah, I noticed that the amount of noise increased with the input resistance. I went with the schematic that was posted earlier in this thread done by the_equalizer, because honestly my knowledge of how triodes work is fairly limited. They don't teach you valves anymore in EE, haha.
 
I guess my next step would be to try to implement some negative feedback though. The sound right now is perfectly acceptable, with almost no noise at normal listening levels unless I'm touching the volume pot (probably a sign that not all of my grounds are done correctly...)
 
Aug 1, 2013 at 11:58 PM Post #6,627 of 7,265

KimLaroux

1000+ Head-Fier
Joined
Aug 2, 2011
Posts
1,090
Likes
69
Quote:
 


Yeah, I noticed that the amount of noise increased with the input resistance. I went with the schematic that was posted earlier in this thread done by the_equalizer, because honestly my knowledge of how triodes work is fairly limited. They don't teach you valves anymore in EE, haha.
 
I guess my next step would be to try to implement some negative feedback though. The sound right now is perfectly acceptable, with almost no noise at normal listening levels unless I'm touching the volume pot (probably a sign that not all of my grounds are done correctly...)

 
Not that hard to do since the amplifier is inverting. Just run a wire from the output to the grid trough a 10 K resistor or so.
 
I'm planning on doing something similar... once I get the damned thing working =/
 
Aug 2, 2013 at 2:37 AM Post #6,628 of 7,265

holland

Headphoneus Supremus
Joined
Jun 15, 2007
Posts
2,210
Likes
22
Quote:
 
Not that hard to do since the amplifier is inverting. Just run a wire from the output to the grid trough a 10 K resistor or so.
 
I'm planning on doing something similar... once I get the damned thing working =/

 
I've done it, not hard as you say, but one would probably not want tons of global negative feedback.  It, IMO, would be better to use a tube with lower mu and reduce the amount of feedback.
 
FWIW, I have the original 19J6, as I got on this when it was published.  I needed local (keep cathode resistor unbypassed) and global feedback (somewhere between 6dB to 12dB), to bring it to a reasonable point.  10K may be too small.  I think I'm running through 100K, and 10K from the pot to the grid.  Injecting after the 10K, to the grid.
 
Aug 7, 2013 at 1:50 PM Post #6,629 of 7,265

Tjj226 Angel

500+ Head-Fier
Joined
Oct 20, 2012
Posts
988
Likes
574
Three last questions guys. 
 
http://www.diyforums.org/SSMH/variants/SSMH-12AU7.gif
 
1: As I look at the schematic for the 12au7 mod, I see that there are two 0.22uf caps in parallel with the 470uf caps. The 0.22uf caps are optional, and with such a little change in capacitance, I am wondering if it would be smarter to buy higher quality 470uf caps, or simply add in the 0.22uf capacitors and leave it alone. 
 
2: Resistors R2 and R8 on the schematic are listed as 390K ohm resistors. However, on the BOM page, they are listed as 220K ohm resistors. Which one should I get? 
 
3: Resistors R15 and R14 look like they just essentially lower the volume a bit. Is this true, or do they serve a greater purpose? 
 
Aug 7, 2013 at 5:50 PM Post #6,630 of 7,265

KimLaroux

1000+ Head-Fier
Joined
Aug 2, 2011
Posts
1,090
Likes
69
Quote:
Three last questions guys. 
 
http://www.diyforums.org/SSMH/variants/SSMH-12AU7.gif
 
1: As I look at the schematic for the 12au7 mod, I see that there are two 0.22uf caps in parallel with the 470uf caps. The 0.22uf caps are optional, and with such a little change in capacitance, I am wondering if it would be smarter to buy higher quality 470uf caps, or simply add in the 0.22uf capacitors and leave it alone. 
 
2: Resistors R2 and R8 on the schematic are listed as 390K ohm resistors. However, on the BOM page, they are listed as 220K ohm resistors. Which one should I get? 
 
3: Resistors R15 and R14 look like they just essentially lower the volume a bit. Is this true, or do they serve a greater purpose? 

 
1: I got Elna Silmic II in there and can't hear any difference with or without the 0.22µf film caps in parallel.
 
2: The BOM is probably for the 19J6. Those resistors were changed in the 12AU7 to lower the heater voltage from 19 to 12 V.
 
3: See http://www.aikenamps.com/InputRes.htm
 

Users who are viewing this thread

Top