patrick,
Please, is there a possible trade off in quality : not matched but still better than something integrated like the opa861 in a listening test ?
What is the challenge in the sorting out VS not sorted out ? Did some sot23 BJT have not close enough spec today to take the non matching risk (in particular if the bjt come from a same batch?
Please, is there a possible trade off in quality : not matched but still better than something integrated like the opa861 in a listening test ?
What is the challenge in the sorting out VS not sorted out ? Did some sot23 BJT have not close enough spec today to take the non matching risk (in particular if the bjt come from a same batch?
I always buy my devices in 100's and match them all.
So I cannot answer your question.
You can try one without matching, measure distortion, and compare with my published results.
Even then we do not know for sure, because we do not know how poorly matched they are.
Or better still, you buy a bunch, measure them all, pick the worst matched ones.
Then build and measure.
Then we all know.
Cheers,
Patrick
So I cannot answer your question.
You can try one without matching, measure distortion, and compare with my published results.
Even then we do not know for sure, because we do not know how poorly matched they are.
Or better still, you buy a bunch, measure them all, pick the worst matched ones.
Then build and measure.
Then we all know.
Cheers,
Patrick
Ok, I can rephrase as I'm not talking about -your-measurement only and not specifically about your design but all those type of design, diamond stage, etc. My question is more basic. What is the goal with matching please ? Noise floor (distorsion?) ? Same gain for the both channel ? Something else ? Distorsion that are so big it is not listenable (crossing between two bjt, etc ?) if not matched or not good enough to be on the level of an oap like the opa861 that share close design but??????
Last edited:
I have the final prices I hope this will get some more interest in getting matched transistors. It would be nice if everyone would at least get 2 channels single so I will not work for free.
- 1 set of PCBs will stay at 20€
- 2 channels single (6 matched pairs) + test (not matched) 20€
- 1 PCB set (12 matched pairs) + test (not matched) 25€
- shipping EU - 10€, World - 14€
SPREADSHEET
GB still open
- 1 set of PCBs will stay at 20€
- 2 channels single (6 matched pairs) + test (not matched) 20€
- 1 PCB set (12 matched pairs) + test (not matched) 25€
- shipping EU - 10€, World - 14€
SPREADSHEET
GB still open
Last edited:
The dual transistors (BTW not those mentioned in post #39, obviously) used in the circuit is 0.40€ each in quantities of 10 at Mouser.
You will need 3 pairs (6 devices) for each channel.
If not group matched, I suggest you purchase ~ double the amount.
That means 30~36 devices in total (2 channels single ended, plus 1 test).
And you will need a SOT23 5- or 6-pin socket for testing.
I assume you already have a transistor tester or multimeter to measure hfe.
Patrick
You will need 3 pairs (6 devices) for each channel.
If not group matched, I suggest you purchase ~ double the amount.
That means 30~36 devices in total (2 channels single ended, plus 1 test).
And you will need a SOT23 5- or 6-pin socket for testing.
I assume you already have a transistor tester or multimeter to measure hfe.
Patrick
cool didactic ressource
. Thanks . So I will try myself between the matched non matched for the hearing decription of what I can hear at my average listening level at home... I'm very curious to experiment the difference level.
Of course one has to match but I tried to know how it is translated in a listening when not matched and if good enough; I mean, distorsion, etc are words and it's always difficult to explain more the feeling of the real life experiment, i.e. listening session.
I don't know if some Toshiba references have closer spec match between transistors, etc, or if some circuitry tipology are supporting less matching : it's far above my head I have not the knowledge understanding nore the tools to measure
Thanks for this, Ripster.
So to be clear, to request a full set of matched transistors for a 2-channel PCB set one needs to enter a 1 in the "2 channels single + test" column and also a 1 in the "additional 6 matched pairs", correct?
> Of course one has to match but I tried to know how it is translated in a listening when not matched and if good enough;
When not matched you have quite a bit more 2nd harmonics.
But as tube people will tell you, 2nd harmonics is honey (or MSG).
So it depends on how much extra honey you want in your soup.
I normally do everything balanced (or super symmetry, if you like).
So I try to cancel even harmonics by 20~30dB.
> I mean, distorsion, etc are words and it's always difficult to explain more the feeling of the real life experiment.
It also depends on how sensitive your ears are.
So I never describe how circuits sounds, as this is purely subjective.
Patrick
When not matched you have quite a bit more 2nd harmonics.
But as tube people will tell you, 2nd harmonics is honey (or MSG).
So it depends on how much extra honey you want in your soup.
I normally do everything balanced (or super symmetry, if you like).
So I try to cancel even harmonics by 20~30dB.
> I mean, distorsion, etc are words and it's always difficult to explain more the feeling of the real life experiment.
It also depends on how sensitive your ears are.
So I never describe how circuits sounds, as this is purely subjective.
Patrick
Ripster 1
Paddy Garcia 1
jimk04 1 UK
markus22ch 1
Vunce 1 USA
John Luckins 1 UK
ixix 1
pfarrell 1 USA
Quantensprung 2
ernesternest 1
Kartk 1 India
rss388 1 USA
androa76 1
Malvin 1 USA
psmith048 1 USA
weissi 2 Austria
Kuka 1 EU
nautibuoy 1 UK
meanie 1 SG
av-trouvaille NL 1 pcb set / 1 pcb set (of 12 matched pairs + non matched test)
___________________
Running total 22 / 1
Paddy Garcia 1
jimk04 1 UK
markus22ch 1
Vunce 1 USA
John Luckins 1 UK
ixix 1
pfarrell 1 USA
Quantensprung 2
ernesternest 1
Kartk 1 India
rss388 1 USA
androa76 1
Malvin 1 USA
psmith048 1 USA
weissi 2 Austria
Kuka 1 EU
nautibuoy 1 UK
meanie 1 SG
av-trouvaille NL 1 pcb set / 1 pcb set (of 12 matched pairs + non matched test)
___________________
Running total 22 / 1
Last edited: