Minimalist integrated buffer for F2 / F2J

Member
Joined 2014
Paid Member
Hi guys, I'm designing PCB for F2/F2J and I thought that before I send it for the manufacturing it might be a good idea to add some minimalistic buffer on PCB, like in F6. I would like to try this amp with Soren R2R DAC connected directly but without the output opamps so the DAC output impedance would be quite high and HF roll-off might happen. But in F2(J) there is only single rail 24VDC and it's pretty "dirty". I have measured the input impedance in LTSpice and it's 4k at 1kHz, but maybe my SJEP model is not accurate. BTW the output impedance shows 3 Ohm 😉

Does anyone have an idea of a good minimalist buffer that can work at "dirty" 24VDC? An output decoupling capacitor is not a problem because F2 has one on the input anyway. I have some 2sk170, 2sj74 (GR, BL, V) but they like higher voltages together.
 

Attachments

  • pcb.png
    pcb.png
    15.5 KB · Views: 370
Last edited:
Thank you, kinku already has for sale boards, but posting to the US only and needs at least 4 orders.

I can produce more and post worldwide but I need to build F2J first, measure it with FFT, check square wave response, check it with my ears to make sure these boards are reliable, but this can take time 😉
 
A little besides the topic, but if you're using the raw output of the R2R network, it might be better to put a buffer right next to the dam1021. Otherwise you should keep the interconnects well shielded and as short as possible. The raw output has an output impedance of 625 Ohm.
 
Thank you, kinku already has for sale boards, but posting to the US only and needs at least 4 orders.

I can produce more and post worldwide but I need to build F2J first, measure it with FFT, check square wave response, check it with my ears to make sure these boards are reliable, but this can take time 😉

Good work.
Just to clarify , one person does not need to order 4 boards but I will need 4 people to sign up and I will keep one. You can take two and some else can do two. 5 is the minimum quantity of fab house. Of course more is welcome and will reduce the cost.
The reason for not shipping outside few of the parts came back recently from US customs and buyer will lose shipping cost entirely.
 
A little besides the topic, but if you're using the raw output of the R2R network, it might be better to put a buffer right next to the dam1021. Otherwise you should keep the interconnects well shielded and as short as possible. The raw output has an output impedance of 625 Ohm.


Good point, but I am using short interconnects with low capacitance, I think no more than 40pF.
 
https://www.firstwatt.com/pdf/art_diy_nutube_preamp.pdf

This shows a choice of J113, 2SK170, LSK170, 2SK370.


Ben, I owe you one :cheers:

This is really cool buffer. On LTSpice I connected it to Mr. Pass PSU CRC 30mF + 5*0.47 in parallel + 30mF loaded at over 2.5A. Then created RCRCRC with 3 x 10 Ohm and 3 x 4700uF, fed with 2Vpp, dummy load of 4k, and got this FFT on the buffer output... attached.


If I am not mistaken this is 81.63 Ohm output impedance.


But I am not going to add the buffer to the main PCB. I will make it on a separate PCB. Maybe there are going to be 3 RCA plugs on the back per channel:
1) buffer input
2) buffer output
3) F2J input

Then I can choose if I want to use buffer or not.

If I want to use buffer I will connect to the source to 1 and short 2 and 3.
Or if I want to use a low output impedance source I will omit 1 and 2 and connect it directly to 3.

sexy!
 

Attachments

  • F2J-ltspice.png
    F2J-ltspice.png
    17 KB · Views: 309
  • fft-buffer.png
    fft-buffer.png
    7.4 KB · Views: 292
Last edited:
I would definitely try it without the buffer. The input of either F2 is about
90 Kohm, so the DAC is not going to have an issue with that unless it is
looking for a very low impedance, which doesn't seem to be the case.

The high frequency crap of the unfiltered DAC is not going to be much of
a problem for a single power fet stage, which is just as likely to serve
as the filter.

If it doesn't float your boat, then put in a buffer, no big deal....

:snail:
 
Thank you Mr. Pass, I will do so.

Do you recommend adding Zobel network (not inside an amplifier but at the back of the speaker) calculated per loudspeaker with F2J? In the simulation, I noticed that there was a really small overshot on the CURRENT plot. When I added Zobel it was nicely rounded. The voltage plot had a much higher overshoot but we don't care about voltage here 😎

When I build this amplifier (and I can't wait for that!) I am going to connect in series with the speaker some small non-inductive resistor like 0R5 and observe on the scope the square wave response :wrench: