• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Interest for potential GB for UDNess PCBs and matched FETs.

This is my first attempt to organize a group buy. I'm offering a lot of transparency re: how I think this could work (or not), so if there are constructive thoughts to make it a better experience, please share.

See Original Thread

UDNeSS, or You don't need Semisouth's

This thread is to gauge interest. I'll also hear any advice or requirements for matching parts. I've offered my thoughts, but I'm pretty flexible. I'm also very new to DIY and electronics in general, so I may be wrong :eek: in some cases.

I would be willing to buy and distribute (in the lower 48 only) a set comprised of -

2 - main board PCBs - using Patrick's Gerbers
2 - supplemental boards allowing use of 2SK209s for Q2, Q3 - using Patrick's Gerbers
4 - (2 matched pairs) Fairchild Mosfets FQA9P25
8 - (2 matched quads) 2SK209BL
2 - J113 with min 8.5 Idss

I'd be willing to match / sort parts for up to 10 sets / 20 channels. I'd be willing to order parts and distribute up to 20 total shipments. If in addition some people wanted boards, but no parts or boards and a pile of FETs to match themselves. Fine by me. I'd base it off interest. To keep it fair, I'd initially limit it to one set of boards and/or parts per person. I'm not going to donate my labor to one person that wants 20 channels. They can sort their own parts. :rofl:

re: MOSFETs matching.

I researched across the forum. There seems to be a lot of discussion around the process to match these parts and parts like them. I would match the Fairchild parts @22V0 / 1A4. Matches within 5% should be achievable with a decent yield. I usually match parts like this without a heatsink after 30s. I'd be willing to use a heatsink jig with clips and take the measurement at 1min. If it's simple and improves reliability / confidence, I'll do it.

If a heatsink is used I will start all parts at roughly the same room temp and take the measurement within a second or two of a given time after the voltage is applied.

I will not:
bolt each part at a standard torque using a specified washer area to deliver the precise thermal coupling;
keep a heatsink at a known temperature;
allow the heatsink to cool between measurements;
use anyone's favorite thermal compound;
use new thermal compound between measurements;
or guarantee timing to the ms.

I'm not willing to go to that level of effort.

In a different field, I've done extensive work determining the precision and accuracy for standardized test methodologies and instrumentation. I've also determined the expected variation within, between, and overall among operators trained in standard methods using standardized samples with standard / calibrated instruments. From what I've seen in the threads - even if I tried to improve the accuracy of the measurement by using heatsinks, it probably would not result in a more precise/reliable match. We're looking for matches, not the absolute value or accuracy of the measurement at operating conditions. We're assuming that if parts match under certain reasonable conditions when measured a certain way - that they will also match under the actual operating conditions closely enough. I actually find it interesting when people record the values to 4 significant digits. My guess based on matching about 100 similar devices, is that the measurement error alone may allow a precision closer to 2. If that's a dealbreaker for some people, no worries. I'm willing to do the work within reason. I also love to learn. So, if there's a measurement method that can be accomplished in <5 mins per part that results in more reliable matching - send it my way with your thoughts as to why it would ensure a better match. I'm not interested in theory re: why it would provide a more accurate absolute result. I understand that already.

I appreciate Patrick's policy of not providing matching data. I will follow suit. My commitment will be to match the parts to an agreed process. Providing values for each part will only serve to cause confusion. IMO, there is just too much potential variability in the measurement methodologies to provide an absolute measurement. If I write down 4 values of 4.44, 4.44, 4.45 and 4.46 - then someone measures them at home and finds... 4.31, 4.32, 4.35, 4.38. I'm not willing to deal with the potential outcry.​

re: 2SK209s

I would match using this jig recommended by Patrick
Peak SOT23 Adapter for Testing Parts on DCA55 DCA75 Model PCA23 Atlas | eBay

It looks nifty. I'll buy it and not pass that cost along. I know I'd use it for other projects. I have a pile of BF862, and I'd have loved to have this jig when matching those! I'd also consider providing 2 matched pents. They're small. There's a reasonable likelihood that someone is going to drop one... damage one etc. There's no way I could send out a replacement part with a known value. OK. I could, but I'd charge A LOT for that. If I get enough parts, and the values match tightly enough, I'd just send along pents vs. quads.

re: J113s
I'd sort all J113s >8.5mA IDSS, then pair them off by the closest values.

Goals
My goal is to build a couple working channels for myself, give a bit of my time back to the community that's been so generous to me, and maybe save a few of you a buck or two and a lot of time.

The first question I anticipate is cost. I have no idea. I've never ordered PCBs. I'd calc the costs as such:

  • PCB cost goes to each person. I'd likely round it to some clean value erring on the side of value for the buyer.
  • Total cost of all the parts gets divided among people that want them. I've been told to expect ~20-30% yield. I'd order about 5x the number of parts on the conservative side. Given the cost discounts at volumes, I think it's still cheaper than buying just the number of parts required. Again, I'd likely round to a simple number erring on the value going to the buyer.
  • I'd ballpark shipping and packaging (based on similar GBs I've participated in) at about $7 per order. I will charge a flat rate just to make it easier for me to collect payments. Don't gripe if your actual shipping + envelope is <$7, and be thankful if yours is > $7.
  • My labor and other parts / jigs I'd buy or build - no cost passed on.
  • The experience of building this fantastic design at such a low cost - priceless :D

I had posted in the other thread that I'd send out all remaining unmatched parts to the group. I was told that was not wise. So, I will follow the advice and not send out unmatched / unused parts. However, like with the 2SK209s if I have enough matches to provide an extra or two, then don't be surprised to find a lagniappe in your goody bag.

Thoughts? Interest?

:cheers: to Patrick (@EUVL), @Morde, and a fantastic DIY community
 
Last edited:

Mark Johnson

Member
Paid Member
2011-05-27 3:27 pm
Silicon Valley
Patrick, this looks like a spectacular project and I am extremely interested to participate. Thank you for floating this "trial balloon" !!

I'm in the lower 48. I pledge to buy the allowed maximum quantity, whatever that turns out to be. I am able to drill and tap my own heatsinks & don't require UMS conformity.

Thank you!
Mark Johnson
 
Last edited:
Non-US AK or HI PM me if interested.

Thank you very much for getting a list started. That makes things much easier for me. Before I order parts, I will confirm a final list by PM. As an example, I think Dennis was potentially being supportive vs. participating in the buy. He's also in Canada :D

I am considering buying:

15 board sets
100 MOSFETs
100 2SKY209GR (Note - I incorrectly had BL in the first post)
50 J113. I was told to expect a high % with Idss > 8.5

This is based on volume pricing discounts and variation in matched set yield. I'll guarantee 10 sets, but we may get lucky on yield. I would use remaining matches for up to 5 extra sets. So, we could have a first-come, first-served waiting list of up to 5. I would calculate final prices based on the number of matched sets I am able to get out. I'll eat the cost of extra boards if they don't sell. Heck, Mark wants as many as possible, so I think we're good :D

Seem fair? Thoughts from the group?

Thanks very much for the interest! I'm looking forward to building this.

Edited to add - If people in AK, HI or non-US countries are interested. PM me. Let me see what I can do. I feel a bit bad leaving people out, but if there's a way to do it easily and successfully - I'm willing.
 
Last edited: