MiniRef Schematic and PCB layout

[linuxguru]

This is a first-pass attempt at building a scaled-down, value-oriented version of a nested gainclone similar to the Mauro Penasa MyRef design. It differs from the MyRef RevC in several ways:

1) LM1875 chipamp as the Howland current pump offers more component flexibility - can be interchanged with TDA2040, 2050, etc.

2) Lower output power of 20W + 20W makes it suitable for bookshelf and HT setups.

3) Two channels/board to reduce overall system costs. The board is only ~15% larger than a single MyRef RevC version 1.3 monoblock.

4) Op-amp flexibility - it will work with numerous commodity (NE5532, etc.) and premium opamps (LM4562, OPA2134, etc.) for the outer voltage-feedback loop, with slight changes in compensation in some cases.

5) Outer loop uses Walt Jung's nested schema, which differs slightly from the MyRef.

6) Compensation schema is much simpler, with a single modified DeBoo integrator per channel, greatly reducing the component count of premium capacitors compared to the RevC.

7) Speaker protection circuit is identical to the MyRef RevA 2-channel schematic.

8) Shared rectifier and PSU caps to reduce board area and costs.

I'm calling it the MyRef Mini because of its close similarity to Mauro's MyRef series. It still needs to be built and validated, but I expect it to be comparable to or better than the MyRef Rev A in audible sonics (with the components shown in the schematic). In audible sonics, it should comfortably crush the Yuanjing and similar EBay gainclones based on the LM3886, LM1875, LM4766, TDA7294, etc., at a slightly higher price point for a 2-channel board.

Here are the exported images of the Schematic and Board Layout (warning: although completed, this is only a first-pass attempt and may have undiscovered errors in both the schematic and layout). The dimensions of the PCB are about 5.5 x 3.2 inches. The BoM should be sub-$50 for the PCB and all the components on the PCB. If there's sufficient interest, I'll eventually have a small batch of bare PCBs fabricated and offered here.

[blakkshepeaudio]

I am interested.

[sasmit]

nice ! but resistor and poly cap solder pads are too small, once the board is made the pad size reduces because of pcb manufacturer drill size being larger than the hole in the component.

[linuxguru]

I believe the pad sizes are adequate, since the parts outlines are from the same library that I'm using with my current MyRef RevC version 1.3 monoblock design, which is completely validated and is currently shipping with no pad or drill size issues. However, I'll eventually re-check all the pads and drill sizes when I go through the CAM output process before handing off for PCB fabrication.

Note that this will be a DS-PTH FR4 board, designed for 35+35 micron Cu. The pads are intrinsically more hardy than single-sided, 17.5 micron Cu boards.

[sasmit]

Originally Posted by linuxguru

I believe the pad sizes are adequate, since the parts outlines are from the same library that I'm using with my current MyRef RevC version 1.3 monoblock design, which is completely validated and is currently shipping with no pad or drill size issues. However, I'll eventually re-check all the pads and drill sizes when I go through the CAM output process before handing off for PCB fabrication.

Note that this will be a DS-PTH FR4 board, designed for 35+35 micron Cu. The pads are intrinsically more hardy than single-sided, 17.5 micron Cu boards.

I had some really hard time with my pcb guy for the first set of boards I made, they too are FR4 PTH but the drill size he used for resistors and small caps was way too large and thus ate away some part of the pad, I had to waste one batch of pcb's before I learnt the lesson

[TheLaw117]

Very nice work! I would never have the patience to actually design one of these boards. Maybe some day in the future.

I think I might try this for my next build.

[AndrewT]

R12=3k3 seems rather high.
Do you have a reason for adopting this value?
RC~0.7us is normal.

[linuxguru]

Originally Posted by sasmit

I had some really hard time with my pcb guy for the first set of boards I made, they too are FR4 PTH but the drill size he used for resistors and small caps was way too large and thus ate away some part of the pad, I had to waste one batch of pcb's before I learnt the lesson

Well, tough luck. That's why you need to use a vendor with CNC drills - just specify the drill sizes in your CAM (Excellon) files and hope that the machines are correctly configured with a full set of drills in the rack.

One of my vendors has exactly the opposite problem - for small batches, he uses manual drilling, ignoring the Excellon files, and usually chooses drill sizes which are about 0.1mm smaller than specified. I now have a small batch of MyRef Rev C boards in which almost all the drill sizes are OK, *except* for a few relating to fast-on tags, which are a bit smaller than specified. I can't make my existing fast-on tags fit, and have to go and hunt for some with thinner legs. Not a show-stopper, but nevertheless an irritant.

[linuxguru]

Originally Posted by AndrewT

R12=3k3 seems rather high.
Do you have a reason for adopting this value?
RC~0.7us is normal.

3k3 and 220pF are exactly the same vales in used in Mauro's Rev C, where it works fine as the input RF filter. That works out to an RC of approx. 0.7 usec, doesn't it? In practice, I sometimes use 2k7 instead, but that's not a significant difference.

[quadtech]

Very nice. Looking forward to the test. I might be interested in a
few, to make a HT amp.
For a multi-channel HT amp setup, it might aid the layout in the chassis
if you could move the LM1875s to the width side of the PCB, kinda
where the LM3886 was located in the MyRef monoblock PCB.