From my reading, the biggest change is to the power supply. Not sure you can do much in that area without the new circuit.
 
I'm sure someone else will chime in.
 
jk

Quote:

boilermakerfan said:

 

If you're chomping at the bit to upgrade, then seek out Toshiba 2238/968s for Q8/Q9 and swap in VitQs for the bypass caps. Then you can always try Silmic II caps as well. 

Click to expand...

What is the benefit of the 2238/968 combo over the 3422/1359s I have? I'm really just curious...I've been out of the loop for so long...

Quote:

tbomb06 said:

What is the benefit of the 2238/968 combo over the 3422/1359s I have? I'm really just curious...I've been out of the loop for so long...

Click to expand...

From Steinchen's site for the 2238/968s:
 
"neutral and balanced sound, tight bass. detailed and clear highs without being bright, easy to listen to
excellent choice but note reversed pinout!"
 
I'll add that the 2238/968 has less bass emphasis but overall better control and a more refined sound.  Better detail resolution or retrieval if your cans can relay that detail.  That's a matter of opinion with Grados. 
 
 
WRT to the PS, you could mod it and it involves cutting a trace to accommodate the polyfuse, the tantalum caps were scrapped too.  It will lower the noise floor, but how much rework do you want to do?  Swap out the tantalum caps with small electrolytics and that will be an improvement. 

Yeah, I don't want to do a ton of rework. It's a real pain to have to switch out a lot of parts, especially the small ones. I'm going to change out the CA4/5 caps from the 470uF to the 1800uF recommended in the new BOM, and change out the CA8 bypass caps to the .18uF VitaminQs, and probably leave it at that. My Max sounds pretty dang good to me as it is, and if I'm going to put a ton of work into it, I'm just going to build a Mosfet Max from scratch and be done with it

Quote:

tbomb06 said:

Yeah, I don't want to do a ton of rework. It's a real pain to have to switch out a lot of parts, especially the small ones. I'm going to change out the CA4/5 caps from the 470uF to the 1800uF recommended in the new BOM, and change out the CA8 bypass caps to the .18uF VitaminQs, and probably leave it at that. My Max sounds pretty dang good to me as it is, and if I'm going to put a ton of work into it, I'm just going to build a Mosfet Max from scratch and be done with it

Click to expand...

That sounds like a great plan.  If I was building from scratch and had 2238/968s I'd probably use them in a different amp, a power amp at higher voltage with big heatsinks so they can really flex their muscles and deliver 10-15W each.  Parallel a set at 15W each for 30W/ch amp, then use a resistor network to protect the headphones and run off the speaker terminals.  Oh wait, I do have the 2238/968s.  Now to find that schematic... 

 
Quote:

tbomb06 said:

Yeah, I don't want to do a ton of rework. It's a real pain to have to switch out a lot of parts, especially the small ones. I'm going to change out the CA4/5 caps from the 470uF to the 1800uF recommended in the new BOM, and change out the CA8 bypass caps to the .18uF VitaminQs, and probably leave it at that. My Max sounds pretty dang good to me as it is, and if I'm going to put a ton of work into it, I'm just going to build a Mosfet Max from scratch and be done with it

Click to expand...

Those sound like reasonable mods to make, if you're really into modding.  The power supply makes the most difference, however.  In the extensive testing that cetoole and I made, though, the biggest difference in the PS was made through an improved ground plane under the PS.  That's kind of hard to put back once the PCB is made without it.
 
The original Millett MAX PCB removed the ground plane from under the PS section to remove some audible ripple and hum in the production MAXes.  However, it left the PS with a performance in the 1-2mV range of ripple.  That's a respectable performance and comparable to an Elpac regulated power supply, but not something we were satisfied with as time went on.
 
Meanwhile, I had finished off one of the very first Millett MAX prototype PCB's - actually a forerunner to the MiniMAX (the very first MAX prototype PCB's were MiniMAX-sized).  When finished with a proposed cut in the ground plane around the PS (very difficult surgery on the PCB), it had a ripple that measured around 0.06 - 0.07mV: a performance that equaled a STEPS or TREAD.
 
That set us on a path of extensive testing to 1) find out the best design for a new MiniMAX PCB (the current one that's been such a success), and 2) find out if the existing MAX PCB's could be sufficiently modded to result in similar performance in the PS.  As things turned out, the bulk of the solution lay in a very small resistance connection between the PS and the amp circuit.  You see, even a TREAD or STEPS usually has wire leads that connect it to an amp.  Those wire leads provide enough of a "buffer" of resistance to keep the LM317 from oscillating, thereby causing the ripple.
 
In the original MAX production PCB's, the connection between PS and amp circuit is a very short, almost zero resistance connection, with a huge bank of capacitors that immediately follow (CA4/CA5 and the Wima bypasses).  Those capacitors are needed in the zero feedback, single-ended circuit to help sink the return ground signal from a pair of headphones, while at the same time helping to supply the current load to the tube heaters and diamond buffer Class A bias.  Anyway, much to our surprise, regardless of the modding - we could not lower the ripple below about 0.2mV. on an existing MAX production PCB.  Conclusion? The ground plane was needed under the PS.
 
That's why the MiniMAX PCB has a different power supply design and the same design was incorporated into the MAX V1.2 PCB as well.  Further testing and tweaking with electrolytics vs tantalums - as BMF stated - lowered the final ripple to around 0.045mV, a truly outstanding performance.
 
This history is chronicled in detail on the MiniMAX website -
http://www.diyforums.org/MiniMAX/MiniMAXhistory3.php

Quote:

tomb said:

 

Those sound like reasonable mods to make, if you're really into modding.  The power supply makes the most difference, however.  In the extensive testing that cetoole and I made, though, the biggest difference in the PS was made through an improved ground plane under the PS.  That's kind of hard to put back once the PCB is made without it.

 
The original Millett MAX PCB removed the ground plane from under the PS section to remove some audible ripple and hum in the production MAXes.  However, it left the PS with a performance in the 1-2mV range of ripple.  That's a respectable performance and comparable to an Elpac regulated power supply, but not something we were satisfied with as time went on.
 
Meanwhile, I had finished off one of the very first Millett MAX prototype PCB's - actually a forerunner to the MiniMAX (the very first MAX prototype PCB's were MiniMAX-sized).  When finished with a proposed cut in the ground plane around the PS (very difficult surgery on the PCB), it had a ripple that measured around 0.06 - 0.07mV: a performance that equaled a STEPS or TREAD.
 
That set us on a path of extensive testing to 1) find out the best design for a new MiniMAX PCB (the current one that's been such a success), and 2) find out if the existing MAX PCB's could be sufficiently modded to result in similar performance in the PS.  As things turned out, the bulk of the solution lay in a very small resistance connection between the PS and the amp circuit.  You see, even a TREAD or STEPS usually has wire leads that connect it to an amp.  Those wire leads provide enough of a "buffer" of resistance to keep the LM317 from oscillating, thereby causing the ripple.
 
In the original MAX production PCB's, the connection between PS and amp circuit is a very short, almost zero resistance connection, with a huge bank of capacitors that immediately follow (CA4/CA5 and the Wima bypasses).  Those capacitors are needed in the zero feedback, single-ended circuit to help sink the return ground signal from a pair of headphones, while at the same time helping to supply the current load to the tube heaters and diamond buffer Class A bias.  Anyway, much to our surprise, regardless of the modding - we could not lower the ripple below about 0.2mV. on an existing MAX production PCB.  Conclusion? The ground plane was needed under the PS.
 
That's why the MiniMAX PCB has a different power supply design and the same design was incorporated into the MAX V1.2 PCB as well.  Further testing and tweaking with electrolytics vs tantalums - as BMF stated - lowered the final ripple to around 0.045mV, a truly outstanding performance.
 
This history is chronicled in detail on the MiniMAX website -
http://www.diyforums.org/MiniMAX/MiniMAXhistory3.php

Click to expand...

Thanks for the info, Tom! I had a feeling you would chime in, since you guided me through building this amp the first time around I also should have known that you would have been determined to take what was already quite good and perfect it So would there be anything I could change to the PS that would be worthwhile? Possibly changing the tantalum caps to equivalent value electrolytics? I'd rather do what I can while I'm making changes so that I don't have to take everything out of the case again...

 
Quote:

tbomb06 said:

Thanks for the info, Tom! I had a feeling you would chime in, since you guided me through building this amp the first time around I also should have known that you would have been determined to take what was already quite good and perfect it So would there be anything I could change to the PS that would be worthwhile? Possibly changing the tantalum caps to equivalent value electrolytics? I'd rather do what I can while I'm making changes so that I don't have to take everything out of the case again...

Click to expand...

Here's the details on converting the original Millett Hybrid MAX PCB to the MiniMAX/MAX V1.2 power supply:
 
First - a partial pic of the original layout.  What you want to cut is the blue trace from C5's "+" pad to the horizontal pink trace it feeds into.  All of this is on the bottom of the PCB, of course.  The trace is all the same, it's just that in the pic, the blue trace turns pink as it goes under the red ground plane.  So, you want to remove it from the C5 "+" pad up to the "T" joint where it goes horizontal across the width of the board.

 
Next, this pic shows the extent of the work. Essentially, the only part to be added is PF1 - a 0.5A polyfuse, something like the one on the MiniMAX BOM - RXE50, or something like that. C5 is deleted, but the existing DR3 must be connected ahead of the polyfuse and so must be twisted from it's original position and soldered to the C5 "+" pad.

 
This should give you a ripple performance of about 0.2mVAC, not as good as a MiniMAX or MAX V1.2 PCB, but almost a magnitude better than the original design.  (As stated before, you need the ground plane under the PS to get any better performance, as in a new V1.2 PCB. )

Awesome info Tom, thanks for taking the time to put that together! So then, should I replace the CR4 10uF 35V tantalum with an equivalent sized electrolytic such as this or should I get the 82uF 35V specified in the new BOM? The modification doesn't seem like a terribly difficult undertaking, and if it makes a big difference I'm sure it's worth doing before I box this thing up for good. If I get the itch to do something else I'll just suck it up and build a Mosfet Max

Quote:

tbomb06 said:

Awesome info Tom, thanks for taking the time to put that together! So then, should I replace the CR4 10uF 35V tantalum with an equivalent sized electrolytic such as this or should I get the 82uF 35V specified in the new BOM? The modification doesn't seem like a terribly difficult undertaking, and if it makes a big difference I'm sure it's worth doing before I box this thing up for good. If I get the itch to do something else I'll just suck it up and build a Mosfet Max

Click to expand...

Yeah, I forgot - CR4 should be the 82uf 35V electrolytic.  Tantalums are much stronger than electrolytics on the uf scale.  So an electrolytic about 10 times the rating of the 10uf tantalum would be about right.  I think when I selected the electrolytic for CR4, it turned out that 100uf wasn't available or else the line spacing was wrong.  82uf is close enough.
 

Ok, I got all my parts in and I'm getting ready to make the aforementioned changes. The only thing I have any concern about now is the actual cutting of the trace from CR5+ to the horizontal trace. Is it necessary to actually remove a chunk of the trace, or will a couple of good cuts across the trace (as in, two parallel cuts) do the trick?

Quote:

tbomb06 said:

Ok, I got all my parts in and I'm getting ready to make the aforementioned changes. The only thing I have any concern about now is the actual cutting of the trace from CR5+ to the horizontal trace. Is it necessary to actually remove a chunk of the trace, or will a couple of good cuts across the trace (as in, two parallel cuts) do the trick?

Click to expand...

You'll find that when you're trying to cut a trace in PCB, it's often quite difficult to tell whether there are tiny imbedded pieces that are still making the connection.  I think it's best to clean out an entire channel.  A good X-acto knife and some patience will result in a good job.
 

Wow, do my ears hurt! I don't think the body was meant to take as much Grado listening as I have been doing over the past few weeks. Those bowls are brutal on the outer ear!





Man this amp sounds good with Grado's! I knew the HF-2 would sound good, but, I was completely caught off guard at how well the sr225 sounds. I was originally *this* close to dumping my 225's as I could never get them to the point of being truly listenable for an extended period of time. This amp has changed that thought completely. This kind of performance makes you understand what the fuss is all about with Grados. While far from accurate, they are a fun listen and really get you into the music. Now I might have to pick up a pair of RS-1's and see how far this can go. Damn you TomB!!





Anyway, I am seriously considering building another one. Use all the same parts in terms of caps, resistors, etc. This time though, build it with BJT's and see if the ability to use larger parts can produce something that might even surpass the minimax in terms of sound quality. My minimax sounds wonderful with the HD650 Maybe I can take them a bit farther using the new board.





My question would be, which BJT's to use? I know some of the toshiba parts are rarer than rare, does anyone know a potential source for something like a 2238/968 combo? BoilermakerFan, where did you source your set (if you do not mind me asking)?





jk

So, I just finished making the changes I had in store for my Max. Hoooooly crap. I don't know if it's just in my head, but it sounds (even more) freakin' amazing now! 
 
It took me a couple of hours to replace the CA4L/R, CA5L/R, CA8L/R, and CR4 caps, along with modifying the power supply. It took a long time to get the polyfuse soldered in, particularly with the DR1 lead attached to the same point. I wound up putting one leg of the polyfuse through the hole, then soldering the leg of DR1 to the top of the board and the leg of the polyfuse. Getting the other leg of the polyfuse in was tricky because they're so thin it kept bending off to one side or the other. Cutting the trace under the board wasn't too bad, it just took a little patience.
 
Anyway, thanks for the advice guys, I really do appreciate it! Now I just need to get my front panel in and get the case finished up and I'll be done with this one... 

Quote:

tbomb06 said:

So, I just finished making the changes I had in store for my Max. Hoooooly crap. I don't know if it's just in my head, but it sounds (even more) freakin' amazing now! 

 
It took me a couple of hours to replace the CA4L/R, CA5L/R, CA8L/R, and CR4 caps, along with modifying the power supply. It took a long time to get the polyfuse soldered in, particularly with the DR1 lead attached to the same point. I wound up putting one leg of the polyfuse through the hole, then soldering the leg of DR1 to the top of the board and the leg of the polyfuse. Getting the other leg of the polyfuse in was tricky because they're so thin it kept bending off to one side or the other. Cutting the trace under the board wasn't too bad, it just took a little patience.
 
Anyway, thanks for the advice guys, I really do appreciate it! Now I just need to get my front panel in and get the case finished up and I'll be done with this one... 

Click to expand...

Glad to hear it made a noticeable improvement.  It kind of lets you know how important it is to have a very clean power supply feeding an amplifier.