Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Millett "Starving Student" hybrid amp
New Posts  All Forums:Forum Nav:

Millett "Starving Student" hybrid amp - Page 412

post #6166 of 7056

Having been convinced that this is the right amp for me to build I have been interested in more fancy power supplies for it. I'm not sure I'm able to design and build one myself (I do a lot of logic design and use almost entirely lm7805s!). If someone can supply me with an easy schematic I'd love to take a look, and then build it on stripboard. (I'll build the amp on stripboard too, but keep the pieces separate).


Or I have also been looking at items like Mouser 709-PLP20-48 along with mouser 693-6200.4315. This seems like an easier answer to me whilst being cheaper than the cisco supplies are in europe. Would the meanwell ps work in Europe (230v 60Hz AC, I believe it does) and well with the starving student?


EDIT: OR even something inline such as Mouser 709-PLN20-48 ? seems really easy and neat compared to the 30€ people seem to want for the ciscos at the moment.


Thanks for any responses and info regarding the subject of powering these amps

Edited by Goobley - 8/23/12 at 3:35pm
post #6167 of 7056

Using a switching LED power supply for an audio amplifier is a very bad idea. If you look at the datasheet for those PSU, you will see that they have 3.8vpp of ripple noise. That's unacceptable in an audio power supply. You need a "Low noise" power supply.


I designed and built a linear regulated PSU for my starving student. If you read the last few pages you should find all the information. It was possible for me because I already had a transformer with an acceptable secondary voltage. Buying one might be cost prohibitive. Actually buying a whole linear regulated 48v PSU might be cheaper. Though I haven't checked Mouser is they stocked a transformer that could be used... I'll do that after dinner. ;)

post #6168 of 7056

It's not as bad as I expected. Mouser stocks economical Hammond transformers that can be used in the PSU I designed. Examples include 546-186E56 at 21$ and 546-186F56 at 29$. Both of these are dual Primary so can be wired for either 115 or 230V.


On the other hand, these transformers alone are the price of a complete switching mode power supply. Add the cost of all the other parts to build a linear PSU, and it's about twice as expensive to go that route. It might not fit well into the "Starving Student" philosophy. Though faced with the choice, I'd build a linear supply, just for the fun of it. Linear PSU are easy to build, so why not? The schematic I posted isn't much more complicated than your average 7805 based PSU. Plus going linear just seems proper.


In post #8422704 I listed a few linear PSU that can be bought pre-assembled. Though they still need either a transformer or a switching PSU, as you can't just feed them main's voltage.


If you prefer a complete, plug-and-play solution from Mouser, I'd suggest 967-LS2548 at 19$. The datasheet says it only has 0.2Vpp of noise, which is good for a SMPS. You can probably just mount if inside your amplifier, and add a power socket on the back.

post #6169 of 7056

How amazingly brilliantly helpful!

I'm really glad I posted :) I'm currently leaning towards the pre-built switchers as I have a slight aversion to dropping more than a few volts off with a linear (needs a lot of heat sinking I'd imagine). Looking at your schematic (I had previously managed to find the pages before and after it!) you just state power diodes so do you mean something like the 1N4002?


Are there any major advantages to the linear other than it looking and feeling right? As I'm not sure I could build something better than 0.2Vpp, then again I have no experience so maybe that's completely wrong.


I also notcied you resistor value of 3017, how accurate does this 48V want to be?

Edited by Goobley - 8/24/12 at 12:16am
post #6170 of 7056



Theoretically, a linear PSU will have a lower noise floor than a switcher. This of course is on paper, and the design and construction of the power supply will have a significant effect on the noise floor. It's possible for a switcher to be quieter than a linear. In my amp, there's obvious noise with my sensitive headphones. The problem is I have no way of verifying if the noise comes from the power supply, or from the tubes themselves. Or if it's just the design of the amplifier that is to blame. Maybe it's too simple?


0.2vpp at 48v is just bellow 0.5%. It might look impressive, but the datasheet of the TL783 states an average noise of 0.003%. That's about 0.0014v at 48v. Now whether the construction of my PSU allows for such a low noise, I'm clueless. I really have to bring my amp to my college's electronic laboratories and try it on an oscilloscope... But as I said, I hear noise on it's output. If it's from the tubes, then 0.5% or 0.003% from the PSU would be irrelevant.


As for efficiency, you are right that it does waste a lot of energy. The TL783 is a old regulator which has a high drop-out voltage, all wasted in heat. I'm estimating the regulator wastes 5W in my amplifier. But if you think this is ugly, maybe this amplifier isn't for you... Each MOSFET drops 5W trough bias current. That's 10W to dissipate from the transistors. It gets very hot.


I did not precise the diodes, as it's not really important. I built it using what I had on hands. I used a 1N400-something as protection diode for the regulator, and a 4A bridge-in-a-cube for the main rectifier bridge. 1N400X diodes are rated for 1A, which is border-line in this application. I feel safer with 4A.

post #6171 of 7056


I'm almost ordering my stuff - will get and use the switching psu for now and then see how it goes. I'm going to try and build the whole thing into one enclosure (psu + amp) and therefore have a bit of mains wiring. (I'm careful enough). How ever I just wanted to know if I should use a double pole switch to turn off live and neutral (leave ground connected and live fused) or just a single pole switch to turn on and off fused live?


Many thanks, this is not my area of expertise at all.

post #6172 of 7056

I don't know how mains are wired in Europe, but here in Canada the Neutral line is actually grounded in the electric panel. This means it's useless to open the Neutral line (N), all you really need is to open the Hot line (L for Live or Line). The problem is, you never know if the main's plug is wired reversed in the wall. It's the first thing I check when I move in a new room/apartment, but that's just me being OCD.


Either way, it's usually not necessary to use a double pole switch to open both lines. In theory, it's safer, but in practice, it's just overkill. I have never seen a media equipment that had a double pole to open both lines. Just use a single pole on the Live line, and place a fuse between the power socket and the switch. Leave Neutral and Ground hard wired.

post #6173 of 7056

I went with a socket for a kettle lead that has a fuse and a switch built in. Everything is now ordered and should arrive within a week. I just need to sit down and start drawing some stripboard diagrams and cutting some wood for the case.

I am thinking of splitting the amp's constituent channels into two seperate boards to give my case a more balanced look, I take it this will not cause any major adverse affects?

post #6174 of 7056

Great! Don't forget to post pictures, they are always welcome.


Someone else built a MSSH on a perfboard recently. It was supposed to be on a stripboard but the builder decided to go with a smaller perfboard in the end. There's pictures and drawings on the thread, it might be useful: http://www.head-fi.org/t/597166/another-millett-starving-student-build


There's no problems with separating the channels. The two channels on my build are pretty much isolated from each others. You can divide C1 and C6 into smaller capacitors, so that you have one C1 and one C6 per board. Same for R13: you'll need one per board. Another option is to build a small board with C1, R13 and C6 on it, and from it send a pair of wires to each channels. I would recommend the first option, since the later misses the point of these capacitors. They are power decoupling capacitors, and so they have to be as close as possible to the item they decouple. In this case it's the MOSFET for C1, and the tubes for C6.

post #6175 of 7056

Would it cause any harm if I didn't change c1 and c6? If I just kept them both at stock values while still splitting to two boards? I did sort of go a bit overboard and buy enough caps and resistors for 2 amps...


EDIT: I have been looking at the power supply I bought, manual is here ( http://us.tdk-lambda.com/lp/ftp/manuals/ls_manual.pdf ) Obviously I connect my circuits inputs to v+ and v-, but do I connect the heatsinks to circuit ground, powersupply ground or v-? I'm inclined to go with circuit ground and therefore logically v-. 


As I am working on some stripboard layouts and would like some advice, should I start my own build log, or keep it contained to this thread?

Edited by Goobley - 8/25/12 at 8:45am
post #6176 of 7056

Yeah no problems, you can use two C1 of the same size, for twice the capacity. Same for C6.


As for the power supply, the manual says the V- and Earth are isolated, more or less. What I'd do is connect all the heat sinks and the enclosure to the PSU's Earth. I'd connect all the amplifier "circuit grounds", or 0V, to V-. If you get some static noise, you can add a 100 Ohms resistor between V- and GND.


You may want to start a new thread if you plan on extensively log your project, with pictures and all. You might get more discussion this way. 

post #6177 of 7056

Kim, that turned out very well. Very sharp build.


Note about original power supply. These are included with Cisco's wireless access points and some VOIP phones. In a building that uses Cisco ethernet switchs the power for these devices comes through the ethernet cable so the little power supplies are general tossed into a box in a data closet, collecting dust.


If you work in a good size office building, school, etc or know someone who does, contact the computer person there. If they are using Cisco devices you might be able to score one for free.

post #6178 of 7056

hello everyone!


Long time reader, first time posting. My question pertains to the 12au7/17ew8 builds.

Since pins 3 and 8 are jumpered, can the cathode cap and 2K resistor be connected to pin 8 instead of 3?

Same question for the input grids on pins 2 and 7? Thanks in advance!

post #6179 of 7056
Originally Posted by Legis31 View Post

hello everyone!


Long time reader, first time posting. My question pertains to the 12au7/17ew8 builds.

Since pins 3 and 8 are jumpered, can the cathode cap and 2K resistor be connected to pin 8 instead of 3?

Same question for the input grids on pins 2 and 7? Thanks in advance!


Yeah no problems! On my build, I actually passed the long resistor's leads trough both solder tabs. Since my build is symmetrical, but the tubes are mounted in parallel, the resistor goes trough 8 to 3 on one side, but from 3 to 8 on the other. I studied a few other builds, and noticed that this is how most people wired this. It doesn't seem to be as important as ground wiring.

post #6180 of 7056

Thanks, Kim for the quick reply! I may end up wiring my cathode caps and resistors just like you did although mine was a space issue cuz  I used coupling and bypass caps rated for 600v since i had them on hand and wanted to keep costs down. the 8-3 and 3-8 method u describe will work perfectly!

New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Millett "Starving Student" hybrid amp