Millett "Starving Student" hybrid amp

[moosefet]

I just signed up to post my build but I am not allowed to post pictures yet

 

[moosefet]

A friend of mine who had also built one of these had a heap of 12SR7s that he decided to ship all the way from NY (I live in Australia) to me for my birthday.
 

 
So far I had only built guitar amplifiers and guitar pedals so this was my first hi-fi project. So easy though!
 
First step was to find a bigass heatsink, I found some cpu for $5 with its heatsink and fan still attached from a second hand shop and grinded away all the bits I didn't need.
 

 
 
I do a lot of acid etching with my pedals and amps so decided to do a design for the top of this
 

 

 
 

 
 
heres a little video of some of the etching in progress: https://www.instagram.com/p/BNniDEPjhxp/
 
I forgot to take a photo of it before I hit it with the colour fill
 

 
 
came out great after wet sanding, you never really know how it will turn out until you get to this step. This was also my first negative coloured etch.
 

 
 

 
 

 
 
didn't dig the silver knob much
 

 
 

 
 
I usually like to be a bit neater than this
 

 
 
First fire up and test was a success B)
 

 
 
then I just added a mammoth 16mm pwr led
 

 
 

 
 
Hope some of you dig it. It sounds brilliant, great job to Peter Millet for the design.

 

[whm3223]

   
Here's a summary I wrote up on another forum:
 

Sorry for more noob idiocy, but is that parts list in the cart from Mouser complete aside from the tubes, tube mounts, PSU, and a box?

 

[Nathan P]

I believe it should be, unless someone messed with the BOM (I've had trouble sharing the wrong link before and had people change my bom thinking they're making their own). If there's something I forgot in there I apologize.
 
Moosefet-that thing looks baddical.

 

[Isophon]

Hi,
 
I have a quick question regarding the 17EW8 version of this amp, I believe 'The equaizer' had discovered this tube as a replacement but back in 2010 couldn't find the spec sheet with the load lines.
 
Tried reading the 400+ pages but this is very confusing.  The actual 17EW8 load line are now available on-line so I wonder if a schematic showing the optimal value has been created.  Would someone with good tube knowledge be willing to find a nice linear bias region with not such a high gain as 50 ? Even a text with the right value found woud be really awesome.
 
Also, has someone come up with a good idea to reduce the very high gain of this tube amplifier ?  The 17EW8 has a mu of 50.  My other headphone amps have a overall gain via global feedback of 5.
 
Data sheet for 17EW8 including load line can be found here :  http://tdsl.duncanamps.com/show.php?des=17EW8
 
The sound is really good but the excessif gain is ruining the whole experience.  My pot has 20 clicks and past the 4th click it's already too loud...
For fun I have included a pic of the HA.
 
Thanks for your help.
Eric
 

 

[tomb]

  Hi,
 
I have a quick question regarding the 17EW8 version of this amp, I believe 'The equaizer' had discovered this tube as a replacement but back in 2010 couldn't find the spec sheet with the load lines.
 
Tried reading the 400+ pages but this is very confusing.  The actual 17EW8 load line are now available on-line so I wonder if a schematic showing the optimal value has been created.  Would someone with good tube knowledge be willing to find a nice linear bias region with not such a high gain as 50 ? Even a text with the right value found woud be really awesome.
 
Also, has someone come up with a good idea to reduce the very high gain of this tube amplifier ?  The 17EW8 has a mu of 50.  My other headphone amps have a overall gain via global feedback of 5.
 
Data sheet for 17EW8 including load line can be found here :  http://tdsl.duncanamps.com/show.php?des=17EW8
 
The sound is really good but the excessif gain is ruining the whole experience.  My pot has 20 clicks and past the 4th click it's already too loud...
For fun I have included a pic of the HA.
 
Thanks for your help.
Eric
 

 
The 19J6 has a lot of gain, too.
 
We took to a practice of implementing input resistors in the SSMH design.  This doesn't necessarily affect gain, per se, but it "pushes" the comfortable volume level out to the midrange of the volume pot travel, lessening the effects of poor channel matching, etc.
 
Most SSMH's used a 50K pot.  So, some multiples of 50K worked well - 50K, 100K, 150K.  These can be standard RN55 resistors, one each in series with the signal input for each channel.  If memory serves, I think most people ended up using a pair of 100K resistors, but try a few values to see what works for you.

 

[Sodacose]

Tried reading the 400+ pages but this is very confusing.  The actual 17EW8 load line are now available on-line so I wonder if a schematic showing the optimal value has been created.  Would someone with good tube knowledge be willing to find a nice linear bias region with not such a high gain as 50 ? Even a text with the right value found woud be really awesome.
 
Data sheet for 17EW8 including load line can be found here :  http://tdsl.duncanamps.com/show.php?des=17EW8

 
Nice build, Eric!
 
Unfortunately the datasheets don't seem to have any plate curves that we might use to easily find a good operating point.  It looks like the plate resistance is around 11k though if judging by the amplification factor and transconductance (Mu = Rp * Gm).  That makes the 33k load resistor on the high side if the two halves are wired in parallel but certainly in the ball park of what we usually want (typically 2-3x Rp, though Rp is halved here because of the parallel tubes). You could try a 15k anode load resistor.  That might reduce gain a bit, but it's always hard to tell what kind of effects changes to the operating point have at such low supply voltages.
 
Rather than adding large resistors on the input, which will create Johnson-Nyquist Thermal Noise, I wonder if anyone has tried experimenting with some negative feedback.  A voltage gain of 50x is a bit over 30db.  According to Peter Baxandall and John Linsley-Hood, negative feedback will increase higher order harmonic distortion up to about 15-20db of feedback, at which point all distortion harmonics begin decreasing linearly. See a good discussion by Nelson Pass here (around Figure 10). So if you can add 20db of feedback or so, you won't suffer the creation of new distortion and you'll end with gain a bit over 10db (which is probably plenty).
 
The tube stage inverts the signal and the MOSFET follower does not. We have an inverted output which means we need a non-inverting input for it be negative feedback. The only place in such a simple circuit would be the grid of the input tube. With some quick back of the envelope calculations, it looks like a 2k resistor in series with the grid (which will also function as a grid stopper, just in case) and a 10k feedback resistor from the output (after the cap) to the junction of the grid and 2k grid stopper would get you in the ball park.  
 
Here's some more useful reading on feedback. Someone may want to check my math!  Getting late here and I don't usually play with feedback much.

 

[henkeman]

I have a bunch of E180F 6688 6J9 tubes that Im wondering if they could work in a Starving Student build? They are pentodes but could be run in triode mode. 

 

[dsavitsk]

Im wondering if they could work in a Starving Student build?

Could they? Sure. Should they? Probably not. You have neither sufficient voltage nor current in the SS application, the heater's need a lot of current and are too low voltage, the mu is too high, and they are tricky to work with.

 

[Gofspar]

Has anyone used a DACT Type 21 Stepped attenuator on their starving students?

 

[tomb]

Yes, but this is sort of the question that Dsavitsk answered - should they have?

The Starving Student is special, but it's because of its niche. A CMoy is special, too, but probably most builders wouldn't put an Alps RK27 in one - not when they could easily build something better.

 

[audiophilenoise]

I've never built an Amp before, but I really fancy this so I'm going to give it a go. A quick question for the techier peeps out there. Earlier in the thread it was said the phono inputs could be replaced by a Jack if prefered. Would it then be possible to do away with it altogether and replace the jack/phono input with this or something like it

https://www.aliexpress.com/item/PCM2704-USB-DAC-to-S-PDIF-Sound-Card-Decoder-Board-3-5mm-Analog-Output-F-PC/32724517687.html?spm=2114.01010208.3.1.UxwpzV&ws_ab_test=searchweb0_0,searchweb201602_2_10065_10068_10000074_10000032_119_10000025_10000029_430_10000028_10060_10000067_10062_10056_10055_10000062_10054_10059_10099_10000022_10000012_10103_10000015_10102_10096_10000018_10000019_10000056_10000059_10052_10053_10107_10050_10106_10051_10000053_10000007_10000050_10084_10083_10000047_10080_10082_10081_10110_10111_10112_10113_10114_10115_10000041_10000044_10078_10079_10000038_429_10073_10000035_10121,searchweb201603_2,afswitch_2,single_sort_3_price_asc&btsid=6922bd54-b764-421e-92da-b398bd722123

by un-soldering the 3.5mm jack on the PCB and soldering wire straight of it and use that as my signal input? In essence I would thereby have created a USB valve headphone amp. Has this been done? I've tried searching but this thread is so big and so technical, its easy to get lost in.

 

[Nathan P]

I've never built an Amp before, but I really fancy this so I'm going to give it a go. A quick question for the techier peeps out there. Earlier in the thread it was said the phono inputs could be replaced by a Jack if prefered. Would it then be possible to do away with it altogether and replace the jack/phono input with this or something like it

https://www.aliexpress.com/item/PCM2704-USB-DAC-to-S-PDIF-Sound-Card-Decoder-Board-3-5mm-Analog-Output-F-PC/32724517687.html?spm=2114.01010208.3.1.UxwpzV&ws_ab_test=searchweb0_0,searchweb201602_2_10065_10068_10000074_10000032_119_10000025_10000029_430_10000028_10060_10000067_10062_10056_10055_10000062_10054_10059_10099_10000022_10000012_10103_10000015_10102_10096_10000018_10000019_10000056_10000059_10052_10053_10107_10050_10106_10051_10000053_10000007_10000050_10084_10083_10000047_10080_10082_10081_10110_10111_10112_10113_10114_10115_10000041_10000044_10078_10079_10000038_429_10073_10000035_10121,searchweb201603_2,afswitch_2,single_sort_3_price_asc&btsid=6922bd54-b764-421e-92da-b398bd722123

by un-soldering the 3.5mm jack on the PCB and soldering wire straight of it and use that as my signal input? In essence I would thereby have created a USB valve headphone amp. Has this been done? I've tried searching but this thread is so big and so technical, its easy to get lost in.

Definitely! If you tie the ground of the USB circuit to that of the amp you should avoid the DAC frying issues as well, though I'm unsure if that could cause noise via ground loops. I'm sure there's a way to do it that wouldn't. I personally might look into something like the grubDAC/pupDAC or one of the hifimediy boards vs the AliExpress example as those are known quantities. It'd suck to get bad sound due to the DAC. An integrated odac would be awesome but spendy.

Another option would be to get the raspberry pi board from hifimediy and integrate that with a wireless connection. Add volumio: https://volumio.org and all you have to do is power it, plug in your headphones, and control it wirelessly from your phone.

 

[rmhuntley]

Following along here and decided to give this build a try. I'm going to swap out the RCA jacks for a single 3.5mm jack as it will be mostly sitting on my desk at work.  one thing I want to do but can't really find a good answer for. I'd like to install a single VU meter. I was looking at this meter. would I also need to incorporate this driver board?

 

[Gofspar]

By any chance do you guys think I could skate by with a 20Kohm stepped attenuator vs a 50k.
 
It's a spare $100 attenuator I have laying around and was wondering if I get by with it.