@schultzsch 4pcs
@codyt 6pcs
@cwlo 4pcs
@Paul M 4pcs
@HelmutHolz 4pcs
@manniraj 10pcs
@aud1out 4pcs
@stablgr 4pcs
@ElliotA 4pcs
@chromenuts8pcs
@Analogico 4pcs
@flocchini 8pcs
@enochRoot 4pcs
@barnstead 4pcs
@rhthatcher 8pcs
@pfdavid 4pcs
@Rodeodave 4pcs
@Lvandoorn 8pcs
@Vunce 4pcs
@ErikdeBest 8pcs
@Robin De Wolf 4 pcs
@PierreQuiRoule 4 pcs
@Botte 4 pcs
@mbrennwa 20 pcs
@jotom750 8 pcs
@SPWONG 4 pcs
@ronex 4 pcs
@woofertester 10 pcs
@thirdicomplex 4 PCs
@urien 4pcs
@ACnotDC 4 pcs
@integra 4 pcs
@maximax77 4pcs
@lcsaba 8pcs
@obh 10pcs
@mazzer 4pcs
@vanofmonks 4pcs
@atiq19 6pcs
@Sidsmith 4pcs
@cygnus x1 4pcs
@A Jedi 4 pcs
@spiggs 4 pcs
Uh…230?
@codyt 6pcs
@cwlo 4pcs
@Paul M 4pcs
@HelmutHolz 4pcs
@manniraj 10pcs
@aud1out 4pcs
@stablgr 4pcs
@ElliotA 4pcs
@chromenuts8pcs
@Analogico 4pcs
@flocchini 8pcs
@enochRoot 4pcs
@barnstead 4pcs
@rhthatcher 8pcs
@pfdavid 4pcs
@Rodeodave 4pcs
@Lvandoorn 8pcs
@Vunce 4pcs
@ErikdeBest 8pcs
@Robin De Wolf 4 pcs
@PierreQuiRoule 4 pcs
@Botte 4 pcs
@mbrennwa 20 pcs
@jotom750 8 pcs
@SPWONG 4 pcs
@ronex 4 pcs
@woofertester 10 pcs
@thirdicomplex 4 PCs
@urien 4pcs
@ACnotDC 4 pcs
@integra 4 pcs
@maximax77 4pcs
@lcsaba 8pcs
@obh 10pcs
@mazzer 4pcs
@vanofmonks 4pcs
@atiq19 6pcs
@Sidsmith 4pcs
@cygnus x1 4pcs
@A Jedi 4 pcs
@spiggs 4 pcs
Uh…230?
@mbrennwa the shipping from China is between 40-60$ to me for example with dhl/fedex. If I go and order alone I will get my 4pcs unmatched.
If we send them to be matched and then to the redistribution and to us the shipping price will be similar but we will get the pcs matched.
Another thing to mention is that I don’t know if the chinese seller will accept to send a few pcs here and there so we will need a forwarder anyway.
@mkc I initially spoken for 100pcs and got a confirmation for the price but not also for the stock. From there it escalated very quickly to 200 and now it goes to 300 but I am mentioning it again I am not even sure if there are 100.
I have no problem as long as there is stock.
The chinese guy is very busy usually and doesn’t reply very often and to mention also that he waits for the replies from a component warehouse so it can need a bit of patience and also hope that there will be stock.
If we send them to be matched and then to the redistribution and to us the shipping price will be similar but we will get the pcs matched.
Another thing to mention is that I don’t know if the chinese seller will accept to send a few pcs here and there so we will need a forwarder anyway.
@mkc I initially spoken for 100pcs and got a confirmation for the price but not also for the stock. From there it escalated very quickly to 200 and now it goes to 300 but I am mentioning it again I am not even sure if there are 100.
I have no problem as long as there is stock.
The chinese guy is very busy usually and doesn’t reply very often and to mention also that he waits for the replies from a component warehouse so it can need a bit of patience and also hope that there will be stock.
I have just received the reply from the seller and seems that there was an error at the warehouse and the stock is 0.
Attachments
well, it was supernice as long it lasted
🙂
don't fret Boyz, Schaded Mosfets aren't that much worse ....... anyway, building joy is what counts, not amp as outcome
Life is still good, with or without SITs
🙂
don't fret Boyz, Schaded Mosfets aren't that much worse ....... anyway, building joy is what counts, not amp as outcome
Life is still good, with or without SITs
What is pras saying?well, it was supernice as long it lasted
🙂
don't fret Boyz, Schaded Mosfets aren't that much worse ....... anyway, building joy is what counts, not amp as outcome
Life is still good, with or without SITs
In the end it's not up to me to decide how you want to go about the testing and matching, and it does not matter much to me anyway (I will curve trace my parts anyway). Still, my two cents on matching: I don't really care too much about matching of a single parameter at one fixed DC operating point (for example: matching of Vgs at a given Vds/Id operating point). To me, the curve shapes are way more important because the curves will tell me much more about how the parts operate with AC/audio signals. I have curve traced about 500 of my LD1014 parts. It takes time, but it's possible.
What does matter to me is that you should make the whole group-buy process efficient and easy, with as little shipping and customs involved along the way.
(whoops, I missed the last page with the latest posts suggesting that this might not happen after all...)
What does matter to me is that you should make the whole group-buy process efficient and easy, with as little shipping and customs involved along the way.
(whoops, I missed the last page with the latest posts suggesting that this might not happen after all...)
Last edited:
Exactly...and perhaps we will have better luck next time. A hiding palette/box of VFET or Germanium extraordinaires could arise at anytime (remember those Sony parts?)well, it was supernice as long it lasted
🙂
don't fret Boyz, Schaded Mosfets aren't that much worse ....... anyway, building joy is what counts, not amp as outcome
Life is still good, with or without SITs
Not sure if this is a just a dream or option but not to long ago when I was working in the aviation industry, we need an IC that was discontinued. Newer ones were to fast and without rewriting the software to add all kinds of wait states and circuit modification we were stuck with servicing what was already out in the field as is. We went to Harris semiconductor and they purchased the die built what we needed. Is was only a 500 piece order. Not sure if they or someone like Linear Systems would take on this order? Yes there was a purchase die/setup fee on top of the 500 piece order. Just my 2cents...dreaming again. Fun to daydream.
If you have an automated way to curve trace 500 pcs, and automatically sort them into matched pairs, I would be interested to know how you do it.
If I recall correctly, Mr. Pass uses his design bias conditions to sort devices into matched pairs.
I will be doing a large quantity matching and sorting exercise soon. After I have the parts in pairs, I will do some curve tracing of the pairs to see how close the families of curves are. My hypothesis that the curves will be close.
Depends on what you mean by "automated".
The curve tracer is fully automated by the computer / software, including the temperature control of the tested part. The temperature control is done with with heater elements attached to a block of copper, which acts as a heat buffer. The copper block has a "quick clamp" to press the part to the metal. The "real" work is to clamp the part on the temperature control block and connect the test leads to the pins of the part. After the test I need to remove the part from the clamp and test leads, put a ID sticker/label on the part or put it into a labelled bag. Then repeat with the next part...
Here are some posts with photos of my curve tracer:
https://www.diyaudio.com/community/...tracer-good-or-not.344199/page-2#post-6121016
https://www.diyaudio.com/community/...tracer-good-or-not.344199/page-4#post-6934950
My experience with the LD1014 and THF51 is that parts with matched Vgs values do not necessarily have similar curves.
The matching is done in two steps (automated in the software):
1. Find pairs with similar Vgs
2. Plot curves of those pairs
By looking at the plots it's easy to find good matches for a given application.
See here for how I look at the curves:
https://www.diyaudio.com/community/...-by-vgs-and-curve-tracing.381338/post-6902977
Automated would be if you run each part through your test rig only once. Either a human or some software will match the pairs with the data generated by the test rig. If you only curve traced and you stored all of the curve trace data, you could compare every possible pair by subtracting all of the curves from PartA minus PartB and sum all of the differences for every pair. Then sort all of the pairs' sums from least to greatest. It would be an interesting project. More work than I have time for at the moment.
In a population of 500 parts, there are 4950 unique pairs. You would calculate all 4950 curve trace subtractions. Then rank the best 500 out of the 4950. If you have a fully automated curve tracer with data collection capability, this could be done.
This is DIY. Whatever makes you happy is cool with me. I am fine with how I do it. I try to do it the way Mr Pass has described his process. I do not do the at-temperature part. That is just a time issue for me.
Ok, that's what I have.
I have done exactly that. However, just looking at the (squared) sum of the differences between the curves of two parts turned out to be misleading. Take a look at these two curve pairs:
(1) https://www.diyaudio.com/community/attachments/2_different_rout-jpg.1013041/
(2) https://www.diyaudio.com/community/attachments/3_different_bias-jpg.1013042/
The first pair would exhibit a smaller (squared) sum of curve differences than the second pair. However, the second pair is clearly a better match if you ask me, and it could be tuned to perfection with a simple twist of the bias pot.
Looking at curve pairs by eye is more useful than trying to calculate some (simple/misleading) number to express the quality of the curve match. To avoid looking at a gazillion of curve pairs I first filter the reasonable matches by their Vgs value. Only parts with similar Vgs are paired for plotting, which removes the poor matches and dramatically reduces the number of plots to look at. This is all done with the PyPSUcurvetrace tools.
Time needed to build a copper heater block with regulated heater elements, or time for slower curve tracing? I can relate to the former. The latter is not an issue if the copper block is large enough to keep the temperature stable and the heater elements are powerful enough to control the temperature. My heater block is rock stable within less than about 0.5 deg.C during testing and does not introduce any delays in the curve tracing.
Last edited:
In your example, there are only two curves and one of the curves is where the device is off. I would want to see 5 or 7 curves in the family and I would want the lowest curve to be more on than off.
Second, it looks like you have only 5 or 6 points plotted in the curve. I find that 50-ish points makes a decent curve that shows threshold voltage.
If your pair of devices requires different Vgs for each device, could that make the channels clip differently?
You are doing what I am doing and then curve tracing to get a qualitative match. It introduces a subjective human element. If you cannot find a formula that repeatably measures what you are matching for, I would rather not do that step myself. Just my preference. In the case of your curve shape preference, you could possibly extract the exponential rise of the curve as the matching parameter, i.e. curve fitting. I think that is what you are doing in your eyeball evaluation.
What I do after finding Vgs at the Vds and Ids of the amplifier idle state, I find transconductance on either side of the idle state and I print that out. I print out low and high transconductance. I sort by more similar transconductance match as well as the Vgs match. My reason is that the Class-A amplifier at low volume is operating at the idle state with a small perturbation in Vgs. This is how I calculate low and high transconductance. It is easy to do once I am at the idle state of the amplifier in my script.
Just to share with you generosity gesture by @yoshida - san
as he informed us, he was able to buy SITs locally from reputable source
I did contact him through PM, to help in arranging buy from same source, ZM Gaijin evidently not up to task to converse in Japanese language
Though, it came to sad fact of named source dried up already, even if their web page still showing stock ( I was fast, but not fast enough)
Anyway, yoshida - san offered to send me substantial chunk of his own personal stash, previously bought, as his personal gift
So, I received package today, along with few pages of Japanese papers inside, so I can practice learning Japanese .......... and now besides language lessons I'll be able to properly pursue usage of funny smallnotsosmall SITs in my concoctions
ZM Happy Camper
as he informed us, he was able to buy SITs locally from reputable source
I did contact him through PM, to help in arranging buy from same source, ZM Gaijin evidently not up to task to converse in Japanese language
Though, it came to sad fact of named source dried up already, even if their web page still showing stock ( I was fast, but not fast enough)
Anyway, yoshida - san offered to send me substantial chunk of his own personal stash, previously bought, as his personal gift
So, I received package today, along with few pages of Japanese papers inside, so I can practice learning Japanese .......... and now besides language lessons I'll be able to properly pursue usage of funny smallnotsosmall SITs in my concoctions
ZM Happy Camper