After all the craze about the Sony VFET parts (2SJ28, 2SK82, etc.) I had this wild idea of collecting as many of these parts as possible, run them through the curve tracer for testing and matching, and then make them available here on diyAudio. (See here why curve matching may be better than simple Vgs matching.)
I currently have 20 Sony parts which I don't need for my own amps, and there are more on the horizon. The more parts there are, the better for the matching, and the better for those in need of these parts.
My idea of making this work is as follows:
I currently have 20 Sony parts which I don't need for my own amps, and there are more on the horizon. The more parts there are, the better for the matching, and the better for those in need of these parts.
My idea of making this work is as follows:
- People sell their VFET parts to me at a reasonable price (yes, I have some money that I can borrow). Send me a personal message (PM) if you have some Sony VFETs.
- I will also accept previously untested parts, parts that may not work properly, or parts of unknown sources (as long as I am sure I can get a refund for bad parts).
- Once there are enough parts in the collection, people can start getting matched parts. The price will be such that it covers the average price I payed to get the parts plus a small fee that helps to keep me motivated.
- People may also send their parts to me and get matched parts in return (for a small matching fee).
Last edited:
It's a nice idea and I will be watching this with interest.
My concern is whether you will be able to acquire an adequate number of parts (at a 'reasonable' price...whatever that may be)
for the matching. My understanding is that for the 2sk82/2sj28 parts there really is quite a large spread.
My concern is whether you will be able to acquire an adequate number of parts (at a 'reasonable' price...whatever that may be)
for the matching. My understanding is that for the 2sk82/2sj28 parts there really is quite a large spread.
So far the average price was about $100 per part. You may think that this is very expensive, but I have also seen $400 price tags...
Take a look at the curve plots in the attachment. It's interesting to see how the 2SK82 tend to be more consistent than the 2SJ28 (with one clear outlier for the 2SK82). The quality of the matches will get better with more parts.
In the meantime, you can take a look at the data and matching/pairing here (will add more data when available): https://github.com/mbrennwa/Sony_VFET
Attachments
It is a good idea, but it will not work.
I already proposed the same concept (with 2SK170/2SJ74s link) and it never worked out, even if I could provide matched devices from day one.
Your number one problem is the number of devices. The yield of matched devices is of course proportional to the total number of devices. With 20 devices you may possibly have a couple of good pairs (10% yield, 90% rejects). With 100 devices the yield may go up to say 50%, but that will not allow you to provide matched devices for "a small matching fee". This is of course not a problem for you as you can sell the rejects for the price you paid for them and keep (or sell) the matched devices. You would probably need closer to 500 devices if aiming for something like a 90% yield. Remember also that you are proposing to match by curve-tracing which will further reduce the yield.
Finally, if unmatched 0.2$ LD1014 devices can be sold for 10-15$ when matched, members interested in buying matched devices back at a later stage might rightfully wonder what the "small matching fee" will be. Is it 10-15$ per transistor or 50+ times the value of the transistor?
Just my 2c.
I already proposed the same concept (with 2SK170/2SJ74s link) and it never worked out, even if I could provide matched devices from day one.
Your number one problem is the number of devices. The yield of matched devices is of course proportional to the total number of devices. With 20 devices you may possibly have a couple of good pairs (10% yield, 90% rejects). With 100 devices the yield may go up to say 50%, but that will not allow you to provide matched devices for "a small matching fee". This is of course not a problem for you as you can sell the rejects for the price you paid for them and keep (or sell) the matched devices. You would probably need closer to 500 devices if aiming for something like a 90% yield. Remember also that you are proposing to match by curve-tracing which will further reduce the yield.
Finally, if unmatched 0.2$ LD1014 devices can be sold for 10-15$ when matched, members interested in buying matched devices back at a later stage might rightfully wonder what the "small matching fee" will be. Is it 10-15$ per transistor or 50+ times the value of the transistor?
Just my 2c.
Hi NicMac, your point about the number of parts is well taken. Yes, more parts is always better. However, I believe you misinterpreted my idea a bit. I am not suggesting to set some sort of a threshold for "matches" or "rejects". The needs for the matching will be different for different builder and will depend on the amplifier designs. I do not just care about Vgs values, but the curves as a whole. This will allow users to choose those parts that will fit their specific needs. It's not about rejecting parts (except those that are broken or fakes). Also, just the fact that the parts were tested and confirmed working may be useful to some people.
I can see where you are coming from regarding your rant about the matching fee. First of all, please understand that this is not about making money -- I have much more efficient ways to become rich. This is about giving something back to the diyAudio community, from which I have profited a lot during the past 17 years. Secondly, maybe I did not choose the "matching fee" term very well. I will not do the matching as such -- it's up to the users to choose their parts based on the data. However, I believe it's fair to ask for some compensation for my work. Curve tracing is a bit more involved than just getting a single Vgs reading, and the curves tell you a lot more about the part than just a single Vgs number!
But yes, the whole idea may not work out the way I am hoping. Let's see.
I can see where you are coming from regarding your rant about the matching fee. First of all, please understand that this is not about making money -- I have much more efficient ways to become rich. This is about giving something back to the diyAudio community, from which I have profited a lot during the past 17 years. Secondly, maybe I did not choose the "matching fee" term very well. I will not do the matching as such -- it's up to the users to choose their parts based on the data. However, I believe it's fair to ask for some compensation for my work. Curve tracing is a bit more involved than just getting a single Vgs reading, and the curves tell you a lot more about the part than just a single Vgs number!
But yes, the whole idea may not work out the way I am hoping. Let's see.
Not ranting about matching fees. It is indeed fair to ask for some compensation for the effort and people are anyway free to buy or not.
If your objective is to give something back to the diyAudio community, just putting up for sale for a reasonable price the 20 Sony parts you have would work perfectly.
More members able to build amps and less parts collecting dust
For these rare parts the demand largely exceeds the supply and not everybody agree that tight matching is required to build a great amp.
For us curve-tracing and matching nerds 20 devices is anyway a hopelessly low number.... Believe me, I do understand you.
If your objective is to give something back to the diyAudio community, just putting up for sale for a reasonable price the 20 Sony parts you have would work perfectly.
More members able to build amps and less parts collecting dust
For these rare parts the demand largely exceeds the supply and not everybody agree that tight matching is required to build a great amp.
For us curve-tracing and matching nerds 20 devices is anyway a hopelessly low number.... Believe me, I do understand you.
Okey, looks like this was a nice idea, but Dennis and NicMac were right. The boys and girls out there are too greedy, and I did not receive many more parts. There were a few "for sale" offers on the horizon here on diyAudio and elsewhere, but these never materialized once I asked for PayPal or similar means with buyer protection. Oh well...
Anyway, I decided to let the parts go. I will post a corresponding "for sale" ad in the Swap Meet forum here on diyAudio.
Anyway, I decided to let the parts go. I will post a corresponding "for sale" ad in the Swap Meet forum here on diyAudio.
Here's the link to the Swap Meet: https://www.diyaudio.com/community/...tested-with-curve-tracer-and-vgs-data.384300/