Millett "Starving Student" hybrid amp - Page 259

No 60Hz available, so I'll go with the 50Hz or the 70Hz

what heat sinks did you use?

I've never measured my heat sinks, but I can touch them, and pick them up for a few seconds.
They are hot, but certainly not 87C.
(my SSMH is caseless right now >_>)

I used the original heat sinks the pmillet used.

12AU7 version

Hi all:

In the past few weeks I've been working on a Starving Student Hybrid version using 12AU7 tubes. Based on a post by Dsavitsk that a 12AU7 version could be 'done with literally one resistor change', I started by creating a simulation in TINA-TI and had a working virtual 12AU7 SSHM in a short time. Of course, having it work in simulation is one thing; making it work in real life could be quite another.

Having first built a P2P SSHM and then the outstanding PCB version by TomB and Dsavitsk I could use my P2P build for experimentation. It wasn't until this weekend that I could sit down and work on the project. Fortunately I can say it worked, as I'm listening to it as I type this. It's been running for almost three hours now, I've tested it with my iPod, my portable CD player and my laptop's headphone out with no problem at all no clipping or noise. Unfortunately I don't have any testing equipment other than my DMM so I can't provide you with any distortion or frequency response figures.

Before the details of the circuit, a disclaimer: I'm not an electrical engineer, merely an electronics aficionado so While I understand some things about electronic circuits very well, I really don't understand others. In short: take what you're about to read with the required grain of salt.

The required resistor change is R2/R8 = 390 Kohms. This moves the voltage at the MOSFET gate to compensate for the heater voltage dropping from 19 to 12 volts. As with the 12SR7 version by Logistic, you need to have proper heatsinking since the MOSFET's will be running at ~35.4 volts @ 150 mA... let's see that gives us 5.31 watts of disipation! Compare to the 2.85 watts disipated by each MOSFET in the 19J6 version!

Other structural changes are also required since the 12AU7 is a 9-pin tube, thus requiring a larger socket. Additionaly, the 12AU7 pinout is different to the 19J6's and this requires rewiring of the connections to the tube.

An important detail to consider when building this version is that the 12AU7 has separate cathodes for each triode (pins 3 and 8), whereas the 19J6 uses a single cathode for both triodes (pin 7). This means you need to connect pins 3 and 8 together with a jumper, along with the 'standard' jumper to tie the two plates together, and the one to tie the two grids together.

Here's a pinout mapping for the two tubes:

19J6=> 12AU7
pin 1=> pin 6
pin 2=> pin 1
pin 3=> pin 4
pin 4=> pin 5
pin 5=> pin 2
pin 6=> pin 7
pin 7=> pin 3, pin 8

Pin 9 for the 12AU7 is the heater center tap, used when connecting the heater filaments in parallel. In this circuit we want the heaters to be in series so pin 9 must remain unconnected. Here's a not-very-good picture of the socket wiring:



Still, there's something I'd like some help with. As described, I built the circuit using IRF610 MOSFET's and R2/R8 = 390 Kohms. According to the TINA-TI simulation this should put the gate of the MOSFET's at ~16 volts. But in the real circuit I'm measuring 14.6 volts !! I'm sure someone can explain why this is happening and if it's ok. My guess is it's not ok since the gate should be sitting at around 4 volts higher than the drain, which is sitting at 12.6 volts (the heater voltage).

UPDATE 3: Forget the above paragraph. Wrong voltage measurement was caused by a sh%&ty, cheap DMM. Measured with my trusty old Kyoritsu K-200 analog FET VOM, the MOSFET gate sits at 16 volts. So all is well with the circuit

I looked at user Logistic's post for his 12SR7 version and he's using R2/R8 = 330K and R4/R10 = 180K. If someone has built that version it'd be nice if he/she posted the voltage at the MOSFET's gate with respect to ground. Still, I think that lowering the value of R4/R10 is not a good idea since it should theoretically increase the bass rolloff in the RC coupling unless, that is, one incremented the value of C2/C4 accordingly.

As to the sound, I really don't consider myself and audiophile and thus I don't think I'd really be able to describe in audiophile jargon how does the thing sound; I can merely say it sounds quite good with both my AKG 701's and my Denon AH-D1001K . I haven't compared it to the 19J6 version so I can't comment on that either... sorry !

About the only bad thing I can say about this version is that the 12AU7 heater glow is very subdued compared to the glow of the 19J6, so the amp doesn't look quite as nice IMHO. Here's a long exposure shot:



UPDATE: just for kicks I decided to try a couple of old 12AX7 tubes I had lying around. I expected to hear awful noise and distortion but guess what.... it WORKS ! beautiful sound out of the 12AX7 too. This most surely means that one could also plug a 12AT7 into this circuit and it'd work without a hitch.

UPDATE 2: I whipped out a couple of new production Russian made Mullard 12AX7's I have saved for use in the phono stage of my old tube amp (when I get around to buying a turntable) and plugged them into the SSHM. These tubes are low noise and microphonics and have long plates and enclosed heaters so they DO NOT glow at all; but the sound... I might just be fooling myself (very likely) but it seems to me these lend more depth and space to the sound. Try them!

cheers!

the_equalizer,

Now that was a fantastic post. I love seeing those types of contributions. I sort of got a shiver down my spine while reading it.

I am just at the "waiting for parts" stage of my Starving Student amp. Great. Now I have to order some 12au7s, new resistors and sockets.

Cheers to you sir!

Nice work!

390K seems about right to me. My guess is that the PS is running slightly low. To compensate, you can simply lower the R2/R8 value until the heater sits around 12.6V.

Thank you very much for your post Eric1110 ! Please post your impressions of this version as soon as you build it; I'd really like to know if it works for you.

A nice benefit of this version is that you can swap in different tubes for different headphones/earphones (tube rolling !! ). In ascending order of gain you'd have:

12AU7
12AT7
12AX7

I don't have any 12AT7's in my parts box, so I can't comment on them, though they should work ok. As mentioned in my post, I tried the AU and the AX and the jump in gain is certainly noticeable; not an abismal, hard-to-control gain jump, but a definite increase in volume.

Also, I tried a couple of new production Russian made Mullard 12AX7's (that I've been saving for my phono preamp) and they sounded very well (but they DO NOT glow, AT ALL).

Anyway, I'll be waiting to hear from your build!

Thank you for your words Dsavitsk; I really appreciate you posting your opinion about this.

Indeed 390K seemed to work in paper, when calculating the voltage divider biasing, and in the simulator but the real circuit seems to say otherwise.

Here are some voltage measurements from it (all with respect to ground):

Power supply: 47.3 V
Tube plates: 28.5 V
Tube grids: 0 V
Tube cathodes: 1 V
Tube heaters: 12.5 V
MOSFET gate: 14.5 V

So, I don't understand very well how MOSFET biasing works, but I thought that since there's no current flowing in the gate-drain circuit, a simple voltage divider should be able to set the gate-to-ground bias without a problem. Maybe I could measure the current flowing through R2/R8.

Hmm.. it just ocurred to me that I took the measurements using a recently bought cheap DMM, it could be affecting the measurements with it's impedance... Tomorrow after work I'll dig out my old analog FET VOM and measure with it just to confirm. I'll let you know how it goes. Thanks again for your posts.

cheers!

Ok, once I got the idea of the error being in my new cheap DMM I simply couldn't wait until tomorrow.

Yeah, I was right. Measured with my analog FET VOM, the MOSFET's gate sits tight at 16 volts so all is well with the circuit. Sorry for the confusion.

Re-measured with the DMM and it shows 14.43 V. I don't know if it's an impedance problem with the DMM or simply a precision problem but I guess I'll be getting my money back.

cheers!

Hi everyone, i've just finished my first diy project a Starving Student.
Here are some pictures or the "beast"






Is there any test i should run before powering up the amp for the first time?

Congratulations on your build! It looks nice and solid! I'm sure you'll be pleasantly surprised to discover that it also sounds really good

As for tests go, here's what I do:

1.- Switch the amp on with no source or 'phones attached and watch for sparks or smoke. Obviously if there are sparks and/or smoke you must turn the amp off immediately and start debugging.

2.- Wait for the tube filaments to start glowing (this should take 30 seconds at most). If they don't come up you can start debugging the circuit.

3.- Use my multimeter and measure DC voltage at the power supply jack (48V) and at the tube filaments (19 v).

4- Use my multimeter and measure DC voltage at the output jack, both left and right. Repeat the measurement at the RCA jacks. Voltages should be 0 (zero). If they're not zero, you have to debug.

5.- If the amp passes all the previous tests, I'll turn the volume knob all the way down, plug in some cheap earbuds at the output, my portable CD player at the input and play some music.

6.- If all went ok, I'll plug in my nice 'phones and enjoy some cool tunes!


One last tip if you haven't read the rest of the tread. Try not to power the amp on or off while your 'phones are plugged in
cheers!

Thanks for the advice, regarding your last tip, there 's a risk of destroying the hearphone?

"Destroy" is perhaps an extreme term in this case, but the potential for damage is there. It's not so much on the SSMH because of the AC-coupling at the inter-stage capacitors. However, there is the fact that the SSMH will exhibit some DC spikes at the headphone jack while the tubes/output capacitors are charging up. For that reason, you should always plug in your headphones after powering the amp up for 30 seconds or so. The same caution applies for shut off, although there's no waiting required and the chances of offset are less likely and if they occur, are transients that disappear much more quickly.

That's why you see headphone-delay relays on many tube hybrids such as the MAX/MiniMAX/MOSFET-MAX, SOHA II, CTH, etc. However, it's a simple precaution to follow without one and is in keeping with the "Starving Student" principle.

Hey guys,

i have been enjoying my ssha for about 9 months now and I'm going to try my hand at the pcb version. I'm trying to make this one a little more of a 'sleek' version and was wondering if an aluminum case would act as a sufficient heatsink for the mosfets or do they require the finned variety that we've all been using?


thanks in advance,

scott

double post

Scott,

This is what I did in the first one I built and I thought it worked fine. Others tried it and thought that it ran a tad hot. if you want to do it, you probably need a slightly larger case than the small Hammond people are using to provide a little extra metal. But, this is why I put extra mounting holes on the PCB -- so you can still mount it without needing the rails to slide it in to.

great, that's what i was hoping to hear. now to go find a case.

Thanks...again.