[Jazper]

Well I started with the idea of building a cmoy for my girlfriend about 3-4 weeks ago, then went on to decide that she should get something a bit better, and that I should use the opportunity to upgrade what I have.

So, I decided (after a week of tossing together options, and pondering over making her an A47 and such) that I'd build an M3.

The M3 has become something of an obsession, the amount of work I've put into it is rediculous at this point and people have begun to ask me what I'm doing/how I'm doing it/what components I use etc etc etc.. so that's what this thread is going to be about. I don't promise that it'll be interesting, but it'll be factual.

The power supply:

I started out trying to make a basic m3, was going to buy a STEPS with it to begin with, then after long winded conversations on msn with Dakiller we came upon this design:
http://www.diyaudio.com/forums/showt...threadid=56106

I bought a 20VA toroidal from the local electronics store and the necessary parts

I made some substitutions on the parts (LM317, 5500uF Low ESR caps, LED/Resistor -10k on the output, 5w bleeder - wire wound resistor rather than 2w metal film) and used 1w ceramic resistors. I also only used half of the design. When building it I just built it on plain perfboard (just copper circles around small holes)

Lesson 1 : always check your resistor values - when I first powered up I got 33.5v which was way too high, I was aiming for 28.15-28.6v (yes, it needs to be that precise - I'll tell you why later), needless to say much resoldering and calculations later I managed to get about what I wanted, after I put some load on the end of it.

When this worked ok I left it and began speccing out parts for the M3.

All up this cost about $25AUD (without the transformer)

This was about the time I decided ok I want something that is somewhat unique.


The m3, part 1:

First step I made was to decide on the elec capacitors on the m3, I looked at cerafines, nichicon muse, black gates, panasonic fc, and even just plain old nichicon Low ESR. After much pondering and thought I decided that the best value for money overall is the SILMIC IIs in terms of having something that is both boutique and something that most people don't use. You always hear about people using cerafines or black gates, but never SILMICs..

The crucial part here was money, or lack of it, I'm a student who works on and off, so every cent counts, or counted at this point, so I asked amb about using 35v caps throughout the M3, after getting the ok, I ordered them all from http://www.audio-cube.nl/elna.htm

From there I just got the excel file from amb's website and started going through all the EXTRA parts I needed (listing them in the excel file.)

I should mention that I'm quite happy with the choices made in the excel file regarding parts from mouser - all are high quality and the only modifications I've made so far are the alu caps and I'm thinking about possibly changing the polyester caps for polypropylene however I'm not keen on having things off board and doubt having them will make much difference.

At the onset I'd decided I wanted to use as much power as I could get into the amp, I plan to run 28v with relatively high quinescent current on the mosfets (well as per the settings on amb's page). The reason for this is so I can drive most anything from the amp without issue. 28.5 is the key voltage because I wasn't sure whether or not I wanted to roll opamps, I know that I like the opa series chips (637/627) but I might like to roll some AD8610s sometime. In the big M3 construction thread amb states that the voltage drop across the m3 is 2v, so 28.5v puts this at 26.5v just under the max rating(27v) for the AD8610s


This is about the time I started looking at cases

Cases/enclosures:

I was looking for something that was about 200mm x 300mm for about say 5 days straight, but have still come up short. Hammond have something but it cost a fortune to get shipped over here and looked a bit foul, everyone keeps saying put it in a 1u rackmount... anyhow I haven't settled on a case yet, getting anything from par-metal detracts from the exclusivity of this particular M3 so I'll probably look at something more unique, perhaps part wooden, perhaps not, I know I'll have to shield the case if I use wood. Then again if I could find something half decent that wasn't huge I'd be happy.

The only requirement is that it somewhat fit on the top of my computer case, for convenience reasons.


The power supply, take 2
After much deliberation and talking with family members I came to the conclusion that I should look at doing something different, nay special with this M3, so that I might sell it for a higher price later (if I ever wanted to sell it... )

So I started looking into different things I could do to improve the m3's design, I toyed with removing the rail splitter and having a dual rail power supply, using uber boutique resistors (with 0.001% ratings) and other ideas before I came upon (again on heavy consultation with Dakiller) that the best way to improve the design, and see a net gain in quality would be to move away from the whole lm317 power supply design and go with something better, the modified jung super regulator.

At this point I was sick of messing with protoboard, I wanted something that worked and was laid out on a nice pcb, I found one particular type that was on the diy audio forums that was a possible but they were out of stock so I decided to look at what our resident expert in SMD here (Peranders) had in stock.

Dakiller warned me that it'd cost a lot to get a board from him, but I persisted, I like P-A's designs. I settled on his JSR03 design( http://home.swipnet.se/~w-50719/hifi/jsr03/index.html ), inquired, bought and then started talking to him about parts. (More info here: http://www.sjostromaudio.com/phpBB2/viewtopic.php?t=14 - note I use the nick Keljian over there)

A few of the parts were obsolete and needed replacements, and some weren't able to be sourced easily, also I wanted to run higher voltage than the JSR03 components can handle (stock) so a few modifications were necessary (notibly alu capacitor voltage ratings - see bom)

After using his forums I've substituted quite a few parts for parts available at mouser, but the notible ones are the RN60D vishay dale resistors, the vishay/Roederstein polypropylene capacitors, and the transistors (as some were unavailable). The polyprop cap leads may need to be bent a bit to fit but being axial leaded they are easier to fit than the square brick wima caps which would have to be off board.

(note R8, r9 and r7 are 2k15, 681R and 511R respectively to get just over 28v)

The voltage reference given in P-A's stock part list is LM431, he says the LM329 is better, so based on that, that's what I'm going to use.

Now P-A's design is beautiful, but it doesn't include a bridge rectifier or smoothing caps, My default choice for caps is nichicon UPW, but it turns out that panasonic FM are cheaper and I'll be ordering from digikey anyhow so they'll be what I use. I'll put these together with the rectifiers on protoboard. 6x1000uF 50v

I've chosen panasonic FM for a few reasons, the main one is that the price is right(less than nichicon), but also the impendance is lower than the Nichicon UPW. I'm hoping that C4 and C12 on P-A's design can be 150uF (as panasonic FMs don't come in 100uF from digikey, and yeah I could use FC instead but I'd rather not as the impendance is much lower on the FM)

As for rectifier diodes, diyaudio people seem to like the MUR820, which is overkill but cheap enough to use, so I plan to use a reasonable clone of it.

I've also added good hookup wire and silver solder to the list of things to buy because I figure why not go the extra mile when I'm going that far with everything else?

As of today (14-6-05) I haven't made the final order from mouser or digikey, but I've bought everything I need from amb, audio-cube, and peranders. Barring final changes, I may make the huge order tomorrow or the day after.


Anyhow, I'll leave it there for now and keep this thread updated.

The bom I'm working from is attached to this message, it is a bit messy (will tidy it up later), but yeah it's all there, hours and hours and hours and hours of work worth of it. The equations on the second page are for the resistor calculations.

Any questions, I'll do my best to answer.


PS. Thankyou everyone who's helped so far especially P-A and my girlfriend for their patience and Dakiller for all of his valuable advice, much appreciated.

Jaz

 

[Jazper]

changed rectifier diodes for RHRP860 (made by harris) soft recovery, >30ms .. expensive too :S but you get that. Ordered locally at RS

btw it should be noted that I know the input to the regulator will be about 33.5v because of the mistake I made earlier with the first power supply. BUT I'm prepping it for 50v in just in case of voltage spikes

I also know that when set to 28.5v the LM317 will output 29.5-29.7v unloaded. (again based on the other power supply design)

I'm hoping that the inrush current charging the 8000uF worth of caps (straight after the rectifier diodes) won't blow the transformer.. though if it's just a spike it should be ok

150uF should be fine, so I'm using panasonic FM & FC for the regulator

 

[Jazper]

well on completion of a project I usually look back and see what kind of things I could do differently, this project is no exception so here's what I would have done differently (about the electronics)

1. Read peranders' website more thoroughly regarding the JSR03
I missed some valuable tips because I didn't, but the problems have been corrected. Plus I wouldn't have ordered all the parts I did.

2. Checked each component against the schematic diagrams - would have avoided the trimmer problem I had early on

3. Checked lead spacing on all of the parts I changed or intended to change - would have not had to spend more on components.

4. Read datasheets for each of the active components and compared to the original component choices- would have debugged the LM329 problem earlier.

5. Went a little overkill on the VA rating on the transformer, rather than getting the bare minimum - cost was an issue, but at the same time I didn't do enough research into the amount of power the thing would draw, and the voltages I'd require.

6. Connected everything on first powerup - not connecting the led meant when I slipped and shorted Q5 on the m3 board meant that I killed them even though the power was off, the caps still held a charge

7. used browndog adapters from the start - no mucking around with trying to solder on to chip sockets.

8. if I have an LM317 in the design or anything in a TO220 package, expect heat to come off it - would have been helpful so I didn't burn my finger on the d44h11 in the JSR03 (which is running at 70C)

9. got heatsinks with pins for the mosfets, (which I could have done cheaply, locally) so I could solder them to the board

10. used vishay dale 1/8w resistors in the jsr, they are rated to 1/8w but can handle 1/4w at a pinch, so would have been fine.

11. used wima or avx caps for the polyester film caps rather than no brand for the jsr (though I don't think this would have made much of a difference at all)

12. used 6000uF as I originally intended for dc smoothing rather than 8000uF - I've settled on 7000uF but it would make more sense to go less overall, I don't need as much as I have.

13. read about lead spacing of the silmics and looked at other options, they are simply too big for the m3 board, perhaps a slightly lower capacitance would have fit much better. (say 220x9 rather than 330x9?)

14. bought more terminal blocks from mouser - I've used 3 in the design, broken one in making it all, would have been better if I bought about 6 or so, just so I had spares

15. checked capacitors with a bleeder resistor before soldering - sparks are bad mmmkay?

16. Selected the fast switching, soft recovery diodes to begin with.

17. considered the tensile strength of the hookup wire I was buying - the core has snapped several times due to flexibility issues and such

How it sounds

note: I listen to a wide range of music on this thing, everything from greenday to vivaldi, through usher, van halen, to live, to louis armstrong.. so it's an overall view

As for how it sounds, it's fun! - non technical easy listening and fun to listen to, I think that's the best way to describe it. It is well refined for critical listening however it also has a warm smoothness to it, that the dtpakiller amp doesn't have. It is a bit dark with the opa637s, but the AD8610s are a nice match for it (ironically as the opa chips were the ones that I generally prefer, hence the hassles getting them earlier). The AD8610s in comparison give the amp "a shot of adrenaline" and really make it come alive, indeed they are livlier than the more laid back opa637/627 combination. Though I feel the opa combo would be better for classical, so it comes down to music preference as to which is better with this amp.

Warm is a good way to define the amp, it doesn't have the bass slam that I loved from the dtpa, but that's not to say the bass is lacking - overall it's less fatiguing to listen to. Easy listening is another way to describe it, it's smooth, doesn't have the sharp edges I'm used to. The bass boost is subtle, but there.

I've said it's smooth, but don't take that to mean dark, the treble is nice and rounded, almost honey smooth, female vocals are a delight on this amp(definately not piercing). It doesn't feel electronic. It feels fluid, liquid, almost organic in the way it sounds.

I'm no audiophile and definately not an electronics engineer, but those are my thoughts on it.

Jaz

 

[skyskraper]

perhaps one of the thlaudio cases would be the go? i mean a rack (it needs 1ru btw) works for me because i have one on my desk with other gear in it, but i can see how you may now want one.

 

[Jazper]

Quote:

Originally Posted by skyskraper perhaps one of the thlaudio cases would be the go? i mean a rack (it needs 1ru btw) works for me because i have one on my desk with other gear in it, but i can see how you may now want one.

What do you mean it needs 1ru? 1ru is about 450x 480cm, which is huge, no way does it need that.

 

[skyskraper]

doh, morning typo, it needs to be 2ru, as in height. 1ru isnt tall enuff to fit the mosfet's heatsinks in. i believe one rack unit is 44mm.

 

[Jazper]

Quote:

Originally Posted by skyskraper doh, morning typo, it needs to be 2ru, as in height. 1ru isnt tall enuff to fit the mosfet's heatsinks in. i believe one rack unit is 44mm.

These are the 1U dimensions - I assure you that 1.5" heatsinks will fit in there.

Dimensions in mm (all units 482.6mm wide):
1U
Panel Height 43.7
Cabinet Height 38.7
Cabinet Depth 245

 

[skyskraper]

1.5" = 38.1mm
internal space = 38.7mm

lets not forget the thickness of the board and the lead ends sticking out the bottom.

and room for the heatsinks to breath (whats the fun in not biasing above the bare minimum?

)

 

[Jazper]

Quote:

Originally Posted by skyskraper 1.5" = 38.1mm internal space = 38.7mm lets not forget the thickness of the board and the lead ends sticking out the bottom. and room for the heatsinks to breath (whats the fun in not biasing above the bare minimum? )

Spose, but I'm open to cutting the case and having the fets stick out of the top so it's a non issue


Update
I've just ordered the rest of the parts, the amb order arrived in the mail today (thanks amb!)

 

[skyskraper]

hehehe ok have fun, i cut the top of one of these cases once, theyre a pita, i had to go to a friends work to use their equipment, working with steel isnt fun

 

[Jazper]

Quote:

Originally Posted by skyskraper hehehe ok have fun, i cut the top of one of these cases once, theyre a pita, i had to go to a friends work to use their equipment, working with steel isnt fun

A friend of mine has a dremel, I'm sure it won't be too hard. We'll see though.

Note: I'm probably going to use 8x1000uF caps for the power supply rather than 6x due to ordering limits on digikey

 

[skyskraper]

the dremel will get chewed alive, try drill with HSS bits then use HSS cutting wheels on a power drill (proper one, not cordless).

why order from digikey when you can get heaps of caps and stuff from RS and Farnell for less then the cost of order + shipping to aust?

 

[amb]

Quote:

Originally Posted by Jazper Note: I'm probably going to use 8x1000uF caps for the power supply rather than 6x due to ordering limits on digikey

If this is after the regulator, be warned that it's probably too much capacitance for the LM317. On power-up the charge-up current will be very high, and the LM317 would probably shut down (or blow up). At the very least, go with a LM338 regulator, or use less capacitance. If these are intended for the C7 capacitors on the M³ pcb, the D1 diode (1N4001) is not rated high enough to handle the charge-up current from 8000uF of capacitance.

 

[Jazper]

Quote:

Originally Posted by amb If this is after the regulator, be warned that it's probably too much capacitance for the LM317. On power-up the charge-up current will be very high, and the LM317 would probably shut down (or blow up). At the very least, go with a LM338 regulator, or use less capacitance. If these are intended for the C7 capacitors on the M³ pcb, the D1 diode (1N4001) is not rated high enough to handle the charge-up current from 8000uF of capacitance.

Nope, this is before the regulator

thanks for the advice though.

C7 caps are 330uf 35v Silmic II x 9

 

[Jazper]

Quote:

Originally Posted by skyskraper the dremel will get chewed alive, try drill with HSS bits then use HSS cutting wheels on a power drill (proper one, not cordless). why order from digikey when you can get heaps of caps and stuff from RS and Farnell for less then the cost of order + shipping to aust?

cause the chips required for the JSR03 are not available here from RS or farnell.

Most parts are coming from mouser, the power supply caps + extra chips are coming from digikey.