The Holy Grail Follower Output Stage

[e_fortier]

30 subscriber is 120 matched devices.
50% yield means 240 devices to match.
5 minutes each for measurement, another 5 for sorting and packing.
That is 8 hours non-stop.
Plus another hour to select the quads from an excel sheet.
Another hour to post and send notifications.
That is a total of 10 hours.

How deep is your love (for DIY) !!!!!!


Best regards, 🙂
Patrick
(Just sharing my experience)

Since when do people who love DIY add up their time spent on a project…we don’t.
Maybe your comment is in the wrong section of the forum.

 

[EUVL]

Maybe you are volunteering to do the matching for the benefit of all ?

🙂


Patrick

 

[lhquam]

30 subscriber is 120 matched devices.
50% yield means 240 devices to match.
5 minutes each for measurement, another 5 for sorting and packing.
That is 8 hours non-stop.
Plus another hour to select the quads from an excel sheet.
Another hour to post and send notifications.
That is a total of 10 hours.

How deep is your love (for DIY) !!!!!!


Best regards, 🙂
Patrick
(Just sharing my experience)

I hear you.

Your arithmetic was too optimistic: 240 devices * (5min + 5min) per FET = 40 hours.

When I receive another batch of FETs and have a bigger statistical pool to select from, I will have a better idea about the matching requirements.

Based on some A/B listening tests of the amplifier running without global feedback and with what I would call "suboptimal FET pairing", the matching process might be quite simple. Nothing more than measuring one or two parameters and dropping each FET into an appropriate bin based on those parameters. The pairing would be done by picking FETs from bins of NFETs and bins of PFETs based on a simple matching criteria. With this approach it is not necessary to label the FETs and maintain a spreadsheet. The bins do all of the work.

 

[EUVL]

You are right, stupid me.


Happy matching, 🙂
Patrick

 

[Zigi]

gary s - Australia
dBel84 - USA
yoaudio - USA
EIArte - USA
kannan_s - USA
von Ah - USA
tom1066 - USA
ACnotDC - Germany
Arthur-HK
william2001 - USA
ItsAllInMyHead - USA
Integra - Canada
Sonnenschein - Germany
birdbox - USA
eriksson-Sweden
Asanden - The Netherlands
manniraj - USA
Robert8 - USA
mordikai - USA
lucadvt - Italy
aud1out - Canada
brimax - Italy
E_Fortier - Canada
CodyT - USA
ml48 - UK
Zigi - Poland

 

[lhquam]

Here is a stacked histogram of the FQA28N15 and FQA36P15 gm (transconductances) for the relatively small number of FETs I currently have. Dark purple shows the number of PFETs (FQA36P15) and yellow is the number of NFETs (FQA28N15). The x-axis values are gm in Siemens (S) for the PFETS. The NFET values are offset 0.4S (10%) to provide a total of 13 pairs that have gm values within 10% of each other. With the small sample size of only 37 NFETs and 41 PFETs, 13/37=35% yield.

I have 50 more FET pairs on order and due for delivery by Mar. 11. With the 10% gm difference pairing criteria, those new fets will be enough for about 9 stereo channels. Either I order 100 more pairs or I widen the gm difference pairing criteria.

In a future post I will explain the consequences of having a greater difference in gm between the NFETs and PFETs.

 

[lhquam]

I have 150 pairs of FQA28N15 and FQA36P15 FETs on order and shipped with an expected arrival date of March 18.
I have done considerable work on measurement of the limited number of pairs that I currently have to verify repeatability of the procedure.

I am about to finalize the PCB layout and get the prototypes fabricated and tested. Then I will place a bulk order for PCBs to provide to the diyers. I will soon post the schematic and PCB layout for comments.

 

[myleftear]

gary s - Australia
dBel84 - USA
yoaudio - USA
EIArte - USA
kannan_s - USA
von Ah - USA
tom1066 - USA
ACnotDC - Germany
Arthur-HK
william2001 - USA
ItsAllInMyHead - USA
Integra - Canada
Sonnenschein - Germany
birdbox - USA
eriksson-Sweden
Asanden - The Netherlands
manniraj - USA
Robert8 - USA
mordikai - USA
lucadvt - Italy
aud1out - Canada
brimax - Italy
E_Fortier - Canada
CodyT - USA
ml48 - UK
Zigi - Poland
Myleftear - Switzerland

 

[woofertester]

gary s - Australia
dBel84 - USA
yoaudio - USA
EIArte - USA
kannan_s - USA
von Ah - USA
tom1066 - USA
ACnotDC - Germany
Arthur-HK
william2001 - USA
ItsAllInMyHead - USA
Integra - Canada
Sonnenschein - Germany
birdbox - USA
eriksson-Sweden
Asanden - The Netherlands
manniraj - USA
Robert8 - USA
mordikai - USA
lucadvt - Italy
aud1out - Canada
brimax - Italy
E_Fortier - Canada
CodyT - USA
ml48 - UK
Zigi - Poland
Myleftear - Switzerland
woofertester - USA

I can do large batch characterization with 0.1% uncertainty of Vgs-Id, transconductance, etc with individual printouts for each device. Matching is just simple sorting. I have a Python script to automate the process.

 

[wg45]

gary s - Australia
dBel84 - USA
yoaudio - USA
EIArte - USA
kannan_s - USA
von Ah - USA
tom1066 - USA
ACnotDC - Germany
Arthur-HK
william2001 - USA
ItsAllInMyHead - USA
Integra - Canada
Sonnenschein - Germany
birdbox - USA
eriksson-Sweden
Asanden - The Netherlands
manniraj - USA
Robert8 - USA
mordikai - USA
lucadvt - Italy
aud1out - Canada
brimax - Italy
E_Fortier - Canada
CodyT - USA
ml48 - UK
Zigi - Poland
Myleftear - Switzerland
woofertester - USA
wg45- USA

 

[lhquam]

Here is the PCB layout and the schematic. There are many options for resistor values, depending on the desired global feedback level, closed-loop gain, and opamp input impedance.

 

[Rush]

gary s - Australia
dBel84 - USA
yoaudio - USA
EIArte - USA
kannan_s - USA
von Ah - USA
tom1066 - USA
ACnotDC - Germany
Arthur-HK
william2001 - USA
ItsAllInMyHead - USA
Integra - Canada
Sonnenschein - Germany
birdbox - USA
eriksson-Sweden
Asanden - The Netherlands
manniraj - USA
Robert8 - USA
mordikai - USA
lucadvt - Italy
aud1out - Canada
brimax - Italy
E_Fortier - Canada
CodyT - USA
ml48 - UK
Zigi - Poland
Myleftear - Switzerland
woofertester - USA
wg45- USA

Rush USA

 

[geoturbo]

gary s - Australia
dBel84 - USA
yoaudio - USA
EIArte - USA
kannan_s - USA
von Ah - USA
tom1066 - USA
ACnotDC - Germany
Arthur-HK
william2001 - USA
ItsAllInMyHead - USA
Integra - Canada
Sonnenschein - Germany
birdbox - USA
eriksson-Sweden
Asanden - The Netherlands
manniraj - USA
Robert8 - USA
mordikai - USA
lucadvt - Italy
aud1out - Canada
brimax - Italy
E_Fortier - Canada
CodyT - USA
ml48 - UK
Zigi - Poland
Myleftear - Switzerland
woofertester - USA
wg45- USA
Rush USA
geoturbo- Italy

 

[vitalica]

gary s - Australia
dBel84 - SUA
yoaudio - SUA
EIArte - SUA
kannan_s - SUA
von Ah - SUA
tom1066 - SUA
ACnotDC - Germania
Arthur-HK
william2001 - SUA
ItsAllInMyHead - SUA
Integra - Canada
Sonnenschein - Germania
birdbox - SUA
eriksson-Suedia
Asanden - Olanda
manniraj - SUA
Robert8 - SUA
mordikai - SUA
lucadvt - Italia
aud1out - Canada
brimax - Italia
E_Fortier - Canada
CodyT - SUA
ml48 - Marea Britanie
Zigi - Polonia
Myleftear - Elveția
woofertester - SUA
wg45- SUA
Rush SUA
geoturbo- Italia
vitalica-Romania

 

[lhquam]

Good news:

• I received, measured, and sorted 50 each of the FQA28N15 and FQA36P15 FETs. 100 more each are are in transit.
• Four of the revised PCBs have been fabricated, stuffed with parts, and tested. Everything looks good.
• One hundred boards are now on order.
• I do not know if I will have more than 70 FET pairs (35 stereo channels) for them.

 

[e_fortier]

Amazing work, thanks a million for your time and dedication.
So far there are 32 members so if you can get to 35 stereo channels that would be great 👍
P.s. for some reason post 174 shows all USA members from SUA…

 

[lhquam]

P.s. for some reason post 174 shows all USA members from SUA…

In anticipation of a new name for the country? 😱

 

[dBel84]

Amazing work, much thanks and gratitude.. dB

 

[Vunce]

Super work lhquam!

If not on the list, are you still planning to share Gerbers?

 

[lhquam]

Yes. When I have the final BOM available I will post it and the Gerbers.

The PCBs from the offshore fab are so cheap in quantity (<$2 per board including shipping), why why would you fab your own?