| |||||||||||||||||||||||||||
| Moderators: cmoy | Welcome. Please log in. | ||||||||||||||||||||||||||
Registration is required to post a new topic or a reply. 
| |||||||||||||||||||||||||||
DIY Workshop » M³ headphone amplifier (part 2) dougigs  HeadWize Fanatic Joined: Sep. 21, 2005 |  Message [#100] posted on: 06-28-2008 06:51 AM CST (US).   |  |  When I have the Sennheiser 650s plugged in, I wish there was a bass CUT. |
hardnrg Member Joined: Apr. 23, 2008 | Message [#101] posted on: 06-29-2008 05:31 PM CST (US). | | Mmm, I feel like the K701 is a touch anaemic in the bass, I think my ears like equal apparent loudness so I tend to find myself applying a gain curve not too dissimilar to a loudness curve for most of my headphones, just to varying degrees. Also, it's probably because much of the music I listen to is electronic-based and heavily reliant on bass components. I also ordered two OPA637 and three AD8610AR, so I'll have a play around with 637/627, 637/8610, and 8610/8610 and see what happens |
| hardnrg Member Joined: Apr. 23, 2008 | Message [#102] posted on: 07-01-2008 08:46 PM CST (US). | | Well, thanks everyone for all the help along the way, especially Ti for being so patient with my amateur mistakes and bunglings. Thanks also to Duc for the fantastic polycarbonate caps, and Dougigs for the recommendations on the polypropelene caps. The build is finally complete, I'm running 627s at the moment and have two 637s and three 8610s on the way to see how they sound, but that only requires the tiny tack soldering to the SOIC-DIP adapter, so basically nothing lol. Right then, I promised final build pics and here is my amp, the first amp I've ever completed, and I'm quite pleased with the end result both in audio performance and aesthetics.
So here is the HiFi2000 Galaxy Maggiorato GX283 with optional 10mm thick front panel.
I like the look of black equipment, and went for a black and red theme, so I paid the extra to have the panel fill-engraved at Schaeffer while it was being milled.
Part of the milling/cutting was to make insets for the Neutrik headphone socket and the control knobs.
The "hi-fi style" feet turned out to be cheapy plasticy things, but the matt textured finish complements the brushed aluminium panel, and that's the look I was after.
The switch isn't a Bulgin, it's made by Lamptron, and I'm pretty sure it has slightly more travel than a Bulgin. I think it looks awesome and really sets off the black/red theme for me. The blue LED is actually a clear narrow beam LED that I sanded down flat so that it would be flush with the front panel and also be diffused. The blue LED is the "cool down" indicator in the event that the Sigma11 heatsinks get too hot, they run much warmer than the M³ heatsinks so I chose them as the thermal trigger.
I had the rear panel milled as I didn't fancy cutting anodised aluminium with handheld power tools, I didn't have it engraved nor did I label it. I think it's pretty self explanatory lol.
Ok, so I guess I'll start from the power. I have a 25VA 2x18V toroid feeding a Sigma11 set at 36V via a power relay which is triggered by an Epsilon24 which itself is powered by a smaller 2x12V toroid. The Sigma11 of course feeds the M³, but also a simplified Epsilon12 which only serves as a muting delay. I have some large polycarbonate caps on the input courtesy of Duc (Ferrari)
[Edited by hardnrg on 07-01-2008 at 10:41 PM.] |
| Beefy Member Joined: Feb. 29, 2008 | Message [#103] posted on: 07-01-2008 08:56 PM CST (US). | | That is possibly the most unconventional layout I have ever seen..... but great work nonetheless!  Just one issue though. With your S11 set at 36V, you will cook the AD8610s, because their maximum voltage is only 27V. |
| hardnrg Member Joined: Apr. 23, 2008 | Message [#104] posted on: 07-01-2008 09:30 PM CST (US). | | Yes, it was a bit of a challenge to get it all to fit and keep sensible things like the mains power away from the input signal, short lengths on the pot leads, etc... But I wanted a small case for my desk rather than a ~430mm wide stackable separate Oh yes, completely forgot about the 27V for 8610... Maybe I need to run a switch to R10 on the Sigma11 to switch between 27 and 36... I think I might give 27V a go by changing R10... I'm lazy though lol, hopefully I like the 637 L/R / 627 G sound the best and can stick with 36V and not have to swap out the main toroid (I don't think sinking that much voltage would be too clever if I went for 27V output using 2x18V toroid in series, the Sigma11 sinks get pretty warm anyway) [Edited by hardnrg on 07-01-2008 at 10:42 PM.] |
Ferrari HeadWizer Joined: Apr. 29, 2006 | Message [#105] posted on: 07-02-2008 05:42 AM CST (US). | | Congratulation! Good to know that your build is ended up well. All the details on the milled 10mm front panel does resulted in a very nice looking M³ ! Now… enjoy the music!  |
| standardome Member Joined: Jan. 23, 2008 | Message [#106] posted on: 07-16-2008 09:10 AM CST (US). | | I'm thinking about building an M³ with the bassboost, but I'm also interested in adding 2 band EQ's at 2k3 and 7k for my Grado's, like this (don't look at the Q numbers; they're not right): Is this doable pre-amp (like with the bassboost), and if so; what would be the best way to do this without degrading quality too much? [Edited by standardome on 07-18-2008 at 10:40 AM.] |
amb Headphone Council Joined: Apr. 1, 2004 | Message [#107] posted on: 07-16-2008 01:21 PM CST (US). | | standardome, that's a fairly complex equalization curve and is best done with a parametric equalizer. |
| standardome Member Joined: Jan. 23, 2008 | Message [#108] posted on: 07-17-2008 06:33 AM CST (US). | | Well, I don't mind making consessions if it's too complex. The bassboost part can already be made onboard with R3=1KΩ, R4=10KΩ, Rbb=30KΩ and Cbb=0,07µF. Say then if I'd only want 1 band cut at 4k7 hz, with a faily wide Q factor, like this; Could that be implemented in here, so that it would be pre-amp? I'm really dreaming about having an M³ that's focussed on my oversparkly Grado. [Edited by standardome on 07-18-2008 at 10:38 AM.] |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#109] posted on: 07-17-2008 12:40 PM CST (US). | | standardome, you are still going to need a somewhat broad, low-Q notch filter to get that kind of response. Instead of doing it on the M³ board (which isn't laid out with that in mind), why not build it into a separate circuit before the M³? I won't design it for you here, but there are lots of web resources, as well as books that cover active filter design using opamps. |
| standardome Member Joined: Jan. 23, 2008 | Message [#110] posted on: 07-18-2008 10:34 AM CST (US). | | I'm now thinking about using the bass boost as a tilt equalizer by using these components for the bassboost; R3=1KΩ, R4=1KΩ, Rbb=20KΩ, Cbb=0,05µF This would gradually boost anything anything <10k; Is this a good idea you think, or is it better to have a separate circuit that doesn't boost the lower frequencies, but cuts the higher frequencies, like you said? |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#111] posted on: 07-18-2008 11:19 AM CST (US). | | You can try it to see if you like the results. Use SIP sockets for these parts so you could experiment with parts values. At the end of the day, though, rather than messing with equalizing the headphone amp response to overcome deficient headphone frequency response, why not simply get better headphones? |
| UglyJoe Member Joined: Jun. 10, 2007 | Message [#112] posted on: 07-21-2008 11:57 AM CST (US). | | I've been looking over the schematic to the M3, just trying to teach myself more about amplifiers, and I saw something that I don't recall seeing before. What is the purpose of C1 in the amplifier? It looks to me like it's a local feedback loop from the output of the opamp to the inverting input, but I don't really understand what it accomplishes. On one side of the cap would be the full voltage of the opamp output, on the other side the same voltage after attenuation in the main feedback loop. I can't visualize what the cap actually "does" with those voltages. Because the frequency on both poles are theoretically the same then it would look like DC voltage to the cap, right? It seems to me that the cap would just be an open circuit in that case and not accomplish anything. So what does it actually do? |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#113] posted on: 07-21-2008 12:57 PM CST (US). | | UglyJoe, it's a compensation capacitor, to tune the ultrasonic frequency response and phase of the amplifier so that it would be stable and produce the optimum square wave response. |
| UglyJoe Member Joined: Jun. 10, 2007 | Message [#114] posted on: 07-21-2008 05:21 PM CST (US). | | And how does that work? Why is that needed here when it isn't used in a lot of the simpler designs? I think the PPA has a lot of similarities to the M3, and I don't believe that it uses a cap in a local feedback loop in this fashion. I'm having a hard time bending my brain around what it does, because it looks to me like it would only be feeding or sinking current into the inverting input of the opamp. |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#115] posted on: 07-22-2008 12:51 AM CST (US). | | UglyJoe, the reason such a compensation cap is there, is because the opamp has wider bandwidth than the output stage. If we don't use a small value cap to "bypass" the output stage and send ultrasonic negative feedback from the opamp output directly back to the inverting input, as you go up in frequency toward where the output stage bandwidth drops off, the opamp won't be getting as much feedback as it should. The effect is a rising peak in the frequency response up in the MHz region, which could cause oscillations (could be parasitic and mild, or quite severe). This is a simplistic way of describing the issue, in reality it's a bit more complex because it isn't simply just the amount of negative feedback at stake. The nature of active devices (i.e., transistors, mosfets, etc., whether discrete or within an IC) is that there is capacitance across the junctions, which forms low-pass filters to cause the bandwidth limiting. As a filter, there are also phase issues to contend with. The ultrasonic phase deviations could be severe enough to make a negative feedback become somewhat positive, which would then cause ringing or oscillations. When an optimum value of compensation cap is employed, it prevents these from happening. As for why the PPA doesn't have such a cap (actually it does, but only in the ground channel), the reason is that it uses multiloop feedback in the left and right channels. The inner loop provides some (albeit weak) feedback before the output stage. It's not as effective at controlling instabilities as a compensation cap, and is probably why some PPAs tends to become unstable under certain conditions whereas M³ is virtually always stable as a rock. During the M³ development phase, I experimented with multiloop feedback, and this is one of several reasons we abandoned it in favor of a compensation cap. |
miles hand Member Joined: Jun. 28, 2007 | Message [#116] posted on: 08-03-2008 06:47 AM CST (US). | |
Could you tell us where you bought your cable from? I'd like to try it and I'm having a bit of trouble finding your source. Hang on - cancel that request. I think I've found it. Company called VCD? I've read about the cable there before without realising they were actually selling it. [Edited by miles hand on 08-03-2008 at 12:23 PM.] |
| fishski13 Member Joined: Jun. 14, 2008 | Message [#117] posted on: 08-06-2008 10:51 PM CST (US). | | they're alive and sounding spectacular!!! just waiting for my knob from Partspipe to arrive.
re: M3 enclosure venting. will 0.5" holes x 12 (2.4 square inches)on top of the Hammond 1455T2201 enclosure, directly over the heatsinks, be adequate for 100 mV quiescent current (this is currently what i have it set at)? what quiescent current should i shoot for with AKG K701, speced at an impedance of 62 ohms and sens. of 105 dB/mW? PACE |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#118] posted on: 08-07-2008 02:40 AM CST (US). | |
You should also drill some holes on the bottom to allow the air to convect through the case. |
| fishski13 Member Joined: Jun. 14, 2008 | Message [#119] posted on: 08-11-2008 09:46 PM CST (US). | | in situ... 
posted these on the "other fora". BOM is bog standard per AMB's. signal wire is solid core copper Kimber Kable sprinkled with fairy dust one question though: do i need to worry about the output impedence with 5 ft. of 18 awg wire between the sigma11 and M3? the sigma11 will sit on a lower shelf, far away from the M3, and 5 ft. is enough to get me there without stressing the connections. and another...i opted to not drill holes in the bottom of the M3 enclosure - the top doesn't get that warm, but where/how (ambient through a top hole, or heatsink directly) should i measure the temp with my Fluke 179, and what is a kosher number? next up...a beta22 PACE
[Edited by fishski13 on 08-11-2008 at 10:22 PM.] |
.| amb Headphone Council Joined: Apr. 1, 2004 | Message [#120] posted on: 08-11-2008 11:35 PM CST (US). | |
5 ft. of 18AWG wire adds 0.032 ohms of resistance per run, and 2.4uH of inductance (see the wire resistance and inductance links at my DIY audio site). The lower these are, the better. However, the M³ has a large bank of rail capacitance at the amp end, so it basically swamps all that out.
I would shoot for heatsink surface temperature of no higher than 65-70 deg. C. |
| Page 1 2 3 4 5 6 7 8 9 10 11 12 13 14 > >> | Next Page | Prev Topic | Next Topic | ||||
 (remove _nospam_ ) |
© Chu Moy, 2001.