• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

FS: one pair of Nelson Pass A40 PCBs

Member
Joined 2009
Paid Member
The power supply for the A40 is probably best built along the lines of the original article.

Since it is Class A, it draws a lot of current - about 1.5A per rail per channel - all the time. You will need a 250VA or better, 2x22V power transformer for each channel. One possible option in Europe is Amplimo type 78015 or 77015, rated for 300VA. A 25A+, 100V+ bridge rectifier and a pair of large - say 22,000uf @ 50V+ capacitors (if you feel like spending money, consider Mundorf HC type) - will complete the power supply for one channel. If you feel like spending even more, consider CLC filtering with another pair of capacitors and a pair of chokes in between for each channel.
 
Alex, some questions for you.

1. You mention you have added a Zobel network as can be seen on the pcb artwork in post #13, C5 is 0.1uF but I cant read what value/rating the resistor is shown as R16 (Note, R16 on original schematic is an emitter resistor for output transistor Q7)

2. In post #17 you show a photo of the pcb, but the Zobel network components are not on that pcb - so I am assuming that is not the latest correct photo.

3. Also the Zobel network components are not on the Mouser project as far as I can see that you linked.
 
Member
Joined 2009
Paid Member
Gary,

Apologies for confusion here. The new boards have no Zobel network, as it was not implemented on the Nelson Pass’ board design as per the original article. The photos in post #17 are correct.

The confusion stems from the fact that my original boards, made years ago and shown in the photos attached to post #1, did have the zobel network implemented (R16/C5 on the silkscreen; should have been R20/C5). The motivation was that the board for the A75 has it, and there was some space available. For the new board, however, I reverted to the original design.

The shared project at Mouser has the resistor for the zobel network (R20, 10ohm 1W), but it seems the capacitor (C5, 0.1uF 100V) has vanished from the list. Sorry about this; I will update the project.
 
Member
Joined 2009
Paid Member
Q11 pinout error in post #20 schematic

I spotted an error in my schematic in post #20 above - the LM334 current source is connected incorrectly. I am sorry I did not spot this one earlier.

In the schematic from post #20, the junction of D1, R8 and the base of Q3 is connected to the adjustment pin of LM334. It should be connected to the V- terminal - see the attached pictures. Because of that, when using LM334 as Q11 with the PCB, its adjustment pin and V- pin must be crossed. I apologize for this.

The markings "D G S" on the PCB are correct, and if you use a JFET or a depletion mode MOSFET as the current source, you can plug it in according to those markings.

On a positive note, you can use DN2540 depletion mode MOSFET in TO-92 package as Q11 on the PCB. Its pinout is "S G D", so it goes into the PCB with the flat side towards R8, not R9. I will post photos over the next couple of days.
 

Attachments

  • A40-Q11.png
    A40-Q11.png
    5.1 KB · Views: 192
  • LM334 datasheet.png
    LM334 datasheet.png
    5.6 KB · Views: 196
  • A40 pcb.png
    A40 pcb.png
    31.8 KB · Views: 190
  • LM334 pinout.png
    LM334 pinout.png
    5 KB · Views: 183
Member
Joined 2009
Paid Member
Thank you Gary for the suggestion. Here it is:

This schematic corrects the error described in the post #26 above and replaces the schematic attached to post #20.

It also adds information on using LM334, DN2540N3, and J111/112/113 in place of Q11. Note that using a MOSFET or JFET as Q11, you need to determine the value of R9 experimentally, so that it sets the current thorough Q11 at 2..3mA.
 

Attachments

  • a40_sch.pdf
    37 KB · Views: 85
  • a40_sch.png
    a40_sch.png
    189 KB · Views: 233
Member
Joined 2009
Paid Member
I have been asleep for longer than expected :eek: But I did test a bunch of LM334, DN2450 and J113 as current sources in place of Q11.

LM334 produces 2mA with the R9 value of 33ohm, as per the datasheet. Reducing R9 to 22 ohm increases the current to 3mA or so.

DN2540 gives 2mA with R9 around 680ohm, with small variation among a bunch of these MOSFETs from the same bag. Note there can be larger variation between different batches, which may require adjusting R9 value, but then the current is not that critical in this particular application.

J113 gives 2mA with R9 around 800ohm, with the standard value of 750ohm making the current somewhat higher. Again, there is some small variation between JFETs from the same bag, with more expected for transistors from different batches. Note that the J113 is rates at 35V of drain-gate voltage, which is just a small margin below A40's -32V power supply.
 
Member
Joined 2009
Paid Member
Example build

This is an example build of the A40 front end with 2N5551/2N5401 bipolar transistors, a DN2540 as Q11 (R9=680ohm, I(D1,D2)=2mA), RN55 resistors, Nichicon FW capacitors. Tested so far without the output stage ("NFB" connected to "-DRV" to close the feedback loop).
 

Attachments

  • A40 8.png
    A40 8.png
    246.2 KB · Views: 111
  • A40 7.png
    A40 7.png
    239.9 KB · Views: 104
  • A40 6.png
    A40 6.png
    243.5 KB · Views: 93
  • A40 5.png
    A40 5.png
    245.4 KB · Views: 354
  • A40 4.png
    A40 4.png
    250.2 KB · Views: 327
  • A40 3.png
    A40 3.png
    242.9 KB · Views: 367
  • A40 2.png
    A40 2.png
    268.3 KB · Views: 334
  • A40 1.png
    A40 1.png
    244.9 KB · Views: 372
Last edited:
Member
Joined 2009
Paid Member
LTSpice simulation of the A40

Attached is an LTSpice simulation of the A40 amplifier. The models for the output Darlingtons are from ONSemi, everything else comes with LTSpice. R11 had to be adjusted to keep the quiescent current of the output stage at approx. 1.5A as per the original article.

LTSpice can be downloaded from Analog Devices.
 

Attachments

  • A40 LTSpice.zip
    3.1 KB · Views: 53
  • A40 sch.png
    A40 sch.png
    77.3 KB · Views: 201
Attached is an LTSpice simulation of the A40 amplifier.
Awesome for sharing that, I really appreciate it. I love running a build through SPICE before assembly. Very good learning when these things combine...


It has been decades since I built my last set of PASS A40 mono block amplifiers. Covid-19 has slowed and cancelled some of my parts orders from abroad. I have everything on hand to build a dual mono block except the XFMR for each channel's pwr supply and Q11 - J111 in my build version. I think the J111's may show up tomorrow but my power supply transformer order looks like it was cancelled from Europe. I may need to shop local?


These Alexcp PCB's are looking great so far. Thank you Diy Audio!
 

Attachments

  • IMG_20200316_191759.jpg
    IMG_20200316_191759.jpg
    905.4 KB · Views: 178
Member
Joined 2009
Paid Member
As stated in the A40 part substitution article, you can use a 10kOhm or so resistor instead of Q11. The amp will have more noise but otherwise will work the same.
You can substitute this part with a convention N channel JFET which has been tested to act as a current source in the circuit with R9. The value of R9 has been chosen so that the JFET passes about 3 mA or so (not critical). You can also substitute in a monolithic constant current source if you can find one, rated at about this current and 40 volts.
If you can’t find any JFET for the current source, or are impatient while waiting for one to arrive, you can simply use a resistor as a poor man’s current source by putting a 10 Kohm resistor where the JFET goes, connecting between R9 and ground. This will work, but is noisier that a constant current source. This can be improved with an electrolytic capacitor (say 100 uf @ 50V) between the negative rail and the junction between R9 and the 10Kohm resistor, with the positive pin of the capacitor between R9 and the 10K resistor.
 
Here is one channel of Sanken 2SD2389 x 3pcs and 2SB1559 x 3pcs. Gonna run it on +/_25 VDC rails. I'm kind of stuck at home so I guess I will build this thing. I forgot to tap the holes for the thermal cut off when I was machining the threads for the darlingtons, I'll do that tomorrow when I go to work on the other heatsink.


I think the Sanken's will be a great match to the A40?
 

Attachments

  • Heatsink.jpg
    Heatsink.jpg
    962.4 KB · Views: 158
  • 2SB1559.jpg
    2SB1559.jpg
    653.9 KB · Views: 149
Last edited:
I did end up using J111 for Q11 and adjusted the leads accordingly.
My heatsinks are not massive and considering supply line voltage variations, I settled on a bias of just over 1 amp per rail. Should provide near 30 watts-ish of class A from 27 volt DC rails.
R11 thus became two resistors in parallel to yield 1890 ohms-ish.
Each heatsink is idle at 42°C-ish

When it goes in the case with the actual power supplies, then I can take away the 'ish' factor... and hopefully soon.

When I get it all together, I may put a full description and complete build pictures in the Pass department of DIYAUDIO but I must give one final nod to alexcp. Cheers & thanks.
 

Attachments

  • IMG_20200325_140127.jpg
    IMG_20200325_140127.jpg
    813.5 KB · Views: 130
Last edited: