The diyAudio First Watt M2x

We are waiting, pop corn in hand, to know how the different input gain stages are performing! What are the sound characteristics of each and what differs from one to the other?

Please, share your experiences for those of us watching from the audience!

Thanks, best regards,
Rafa.


My experience are the following;


Austin: Sound too warm and too thick for my system, least prefer.

Tucson: Good, better than Austin, it sound like ISHIKAWA but less refine.

Isikawa it's M2 so you know what it's like.

MOUNTAIN_VIEW This is definitely the best, the sound is big, full and still maintain good separation and details and it's runs real hot!

The differences for all input stage are quite obvious, you can notice them immediately.

As always, these are my opinions, YMMV.

My system
Rega Apollo-R
Pre Cat sl1 clone
BA3 clone
6n8s Aikido

Spk Diy 2-ways TL 92db
 
Member
Joined 2011
Paid Member
The Mountain View circuit has no opportunities for device matching. It is after all a single ended class-A amplifier; single ended means there aren't any pairs of devices that work together in push-pull to generate an output.

While I was testing Mountain View on the lab bench, and tweaking its response at high frequencies, I discussed the circuit a couple of times on the phone with John Curl. John slightly prefers using hand-selected J113 JFETs rather than any-old-J112-that-rolls-off-the-manufacturing-line. But I didn't think diyAudio members would be ready, willing, and able to hand-select JFETs so I didn't follow his suggestion. Besides the circuit as-is sounds GREAT, with any old J112 that rolls off the manufacturing line. So the official Mountain View schematic shows J112, the official Mountain View BOM shows J112, and I don't recommend using anything else unless (a) you really know what you're doing; and (b) you're a crazy person willing to do anything for that last microgram of goodness.

If so then select a J113 that satisfies this inequality: (9mA < IDSS < 12mA). Measure at Vds=9V, i.e., use a battery and a DVM. If you don't know what "IDSS" means then it's probably true that you don't meet criterion (a) above :(

I gotta warn you, not very many J113s will fall into this IDSS range. So you may need to buy 100 parts and test many/most of them, to guarantee that you'll find two J113s in this IDSS range. Fortunately Mouser's qty=100 price is only $0.12 per part. These two hand selected JFETs for a stereo pair of Mountain View boards, will cost you less than the Toshibas for a single Ishikawa board. {if the time you spend in the lab measuring IDSS is assumed to cost nothing}

_
 
Thanks for the detailed response Mark. I think I'll stick with the "any old J112 off the line" since the amount of time I would need to match 100 JFETs is a bit hard to come by these days.

Interesting you bring up the J113's. I was just browsing an older thread about replacements for the 2SK170/2SJ74 in the F5 circuit. You might be familiar with this thread already, Mark. ;) The J113's seem to have some merit in this regard. Good reading if you're into that sort of thing. :D

how about using J111 J175 Jfets in F5
 
I have a rookie question regarding the Tucson Boards. In the provided literature Mark states "When substituting, make sure you choose opamps rated for at least 30 volts of total supply (VCC minus VEE)". Does that mean if the DIP-8 or SOIC-8 says +-18, this would fall within the safe guidelines since the total supply would be 36 total supply?
 
Member
Joined 2011
Paid Member
Remember that miserable math class in school where they talked about negative numbers? Here's a chance to finally use them in your everyday life.

If VCC=+18 and VEE=-18 then (VCC) - (VEE) = (+18) - (-18) = +36.

Since 36 > 30, it ought to work fine in a Tucson board. As you expected.
 
Remember that miserable math class in school where they talked about negative numbers? Here's a chance to finally use them in your everyday life.

If VCC=+18 and VEE=-18 then (VCC) - (VEE) = (+18) - (-18) = +36.

Since 36 > 30, it ought to work fine in a Tucson board. As you expected.

Lol my degree is in economics and finance so I’ve had enough math to last me a lifetime. Since I’m new to all this wasnt sure if there was some other aspect I was overlooking
 
Member
Joined 2011
Paid Member
For those who wish to experiment with gold plating the Mountain View input stage, let me gently suggest "Moffett Field" (an airport inside the Mountain View city limits, where Sergey Brin of Google keeps his personal Boeing 767 airplane).

  1. Start with the Mountain View schematic.
  2. Discard Q2 and Q3 and R4 (node "EMIT" disappears)
  3. Connect C1 to node "SOURCE" instead
  4. Replace J1 with three Fairchild J113 (not J112) JFET devices, connected in parallel. Call them "J1a, J1b, J1c".
  5. Hand test and match J1a, J1b, J1c so their IDSS values are within 10%. Any IDSS greater than 7mA will be fine, but make sure all three JFETs have the same IDSS.
  6. Modify the bias network (Q1 + R3 + crud on the left) of constant current source Q3. Change it in such a way that the current in Q3 is a constant 20 milliamps, pulled from the parallel connection of J1a-J1c. I.e. each of the three parallel JFETs carries 6.67mA of bias current. Make this 20mA independent of temperature and independent of supply voltage, to the extent possible.
Interested builders can fill in the details as they see fit. This may be an opportunity to get rid of the much-maligned MV5075C LED and its horrible "do not substitute" requirement in the BOM.

~
 

Attachments

  • Moffett_Field.png
    Moffett_Field.png
    131.2 KB · Views: 1,218
Try the maneuver discussed in posts #201 and #202, it worked for the guy in Portugal.

_

I would like to, but shipping is 18€ since I don't have anything else to order there :D I have all parts in Mouser's shopping basket for main boards + couple front ends also. Maybe I just go with Tuckson at first so I can get things rolling with this, worry about that LED later.
 
Member
Joined 2011
Paid Member
If you would like to re-engineer the Mountain View input stage daughter card, and provide technical support both now and in the years to come, please go right ahead. I suggest you pay careful attention to its performance (i) versus ambient temperature; and (ii) versus total supply voltage (|VCC| + |VEE|), since those are likely to vary from owner to owner. Not everybody uses the AnTek transformer, not everybody has AC mains within tolerance, and not everybody has air conditioning in their listening room.

Start by measuring a V.1 original board; there are some things you just can't get from informed guessing, or from vendor-supplied (crappy) simulation models.
 
Member
Joined 2011
Paid Member
If you're going to engineer an input stage daughter card, you'll need a test fixture. One of the ones I used is shown below. It includes BNC connectors for extremely low inductance ground connections, much lower than a scope probe's ground wire + crocodile clip. This is needed when doing 2 nanosecond risetime square wave testing; any parasitic inductance in the ground leads will ruin the waveshape very quickly! Of course SMB or SMA connectors would have been even better but I didn't feel like spending that kind of money.

I included (socketed) inductors in series with the power supplies, to assess the input stage's propensity to oscillate if the power supply PCB traces (on the M2X mother board) are very long. I also included (socketed) load resistors to see whether the input stage's performance degraded whe driving the ~600 ohm input impedance of the Edcor transformer + output stage components. And I included (socketed) load capacitance too, to find out whether the input stage becomes unstable when driving a capacitive load.

Sockets let you quickly change a component value, or remove it entirely, or replace it by a short circuit. As you begin testing your input stage board(s) I think you'll quickly learn which of the socketed parts has the greatest effect, and which are essentially dont-cares.

_
 

Attachments

  • With.jpg
    With.jpg
    478.2 KB · Views: 978
  • Without.jpg
    Without.jpg
    275.6 KB · Views: 1,591