Jason Stoddard
Sponsor: Schiit Audio
So, the Vidar first article boards are in, and with the exception of one hilarious mistake (and one possible show-stopper problem), they're looking good.
The hilarious mistake: they installed all the through-hole transistors backwards. See photo.
The metal parts of the large transistors should be pointed towards the heatsink, not inwards. That, plus a couple of missed parts = not a huge deal, and pretty typical of first articles. We respond with corrections, the PCB assembly house makes some changes, and production boards start flowing shortly.
But...there's a potential showstopper.
Here's the problem: the output transistors are too large. And yeah, that might sound a bit weird. But to understand why, you have to understand Toshiba's weird part numbering system, where a 2SC5200O is an old, high-beta-bin TO-3P package (which is what we designed for) and the 2SC5200OTU is a TO-264 package (larger than we designed for.) The correct new part number is 2SC5200N, which drops the beta binning, because they are all apparently high-beta-bin parts now, but in the past they would have been the low-beta-bin parts. Argh. What happened is we ordered the 2SC5200O and got 2SC5200OTU (and the PNP complements, of course.)
"So, yeah, get the right parts, so what?" you ask.
Well, the problem is availability. A cursory glance doesn't show a lot of the correct parts in stock. And we need tonnage of them--now, and in the future. If this stocking situation is correct, we have a problem. It's a problem we can address, but I'm hoping that the distributors we work with will simply shrug and say, "Yeah, no worries, we have plenty of those, we're shipping to your assembler today." And that's likely what will happen on Monday, since it's happened many times before. Our distributors are usually pretty great about making sure the parts we want are continuously in stock, or advising us if we're designing ourselves into a shortage nightmare. I've already sent them, and the assembly house, a note explaining the situation and its urgency.
So what happens if they look around on Monday and find out there really aren't any parts? Well, we have two options, because changing the clip to work with the bigger parts isn't an option (it's too close to the parts):
1. We sub out a similar Toshiba part. Luckily, Toshiba makes some similar parts, some of which are in stock at lots of places. How similar? They have exactly the same specs and use exactly the same charts and graphs as the parts we spec'd. Now, the question of why they'd produce several (seemingly) identical parts befuddles me, but if they are really that similar, they should drop in and work. I have some of these coming on Monday and will drop them in and re-qualify. If they work, then we're set. If they need additional compensation/parts tweaks, then it may cause a few days delay (if we have to go this way.)
2. We sub out a similar On Semiconductor part. On Semiconductor also makes similar parts. Not as similar as Toshiba, but they claim to have a drop-in replacement. This is a lot more iffy, because a different manufacturer's parts are, well, gonna behave differently. We'd have to take a long, hard look at them, even if they seemed to work. This might mean a longer delay. Or not. As with the Toshiba alts, I have some coming in Monday to see how they work.
On the plus side, we'll (at least) be getting data on a lot of alternate parts for Vidar (or other possible power amps). It certainly can't hurt in the future!
The hilarious mistake: they installed all the through-hole transistors backwards. See photo.
The metal parts of the large transistors should be pointed towards the heatsink, not inwards. That, plus a couple of missed parts = not a huge deal, and pretty typical of first articles. We respond with corrections, the PCB assembly house makes some changes, and production boards start flowing shortly.
But...there's a potential showstopper.
Here's the problem: the output transistors are too large. And yeah, that might sound a bit weird. But to understand why, you have to understand Toshiba's weird part numbering system, where a 2SC5200O is an old, high-beta-bin TO-3P package (which is what we designed for) and the 2SC5200OTU is a TO-264 package (larger than we designed for.) The correct new part number is 2SC5200N, which drops the beta binning, because they are all apparently high-beta-bin parts now, but in the past they would have been the low-beta-bin parts. Argh. What happened is we ordered the 2SC5200O and got 2SC5200OTU (and the PNP complements, of course.)
"So, yeah, get the right parts, so what?" you ask.
Well, the problem is availability. A cursory glance doesn't show a lot of the correct parts in stock. And we need tonnage of them--now, and in the future. If this stocking situation is correct, we have a problem. It's a problem we can address, but I'm hoping that the distributors we work with will simply shrug and say, "Yeah, no worries, we have plenty of those, we're shipping to your assembler today." And that's likely what will happen on Monday, since it's happened many times before. Our distributors are usually pretty great about making sure the parts we want are continuously in stock, or advising us if we're designing ourselves into a shortage nightmare. I've already sent them, and the assembly house, a note explaining the situation and its urgency.
So what happens if they look around on Monday and find out there really aren't any parts? Well, we have two options, because changing the clip to work with the bigger parts isn't an option (it's too close to the parts):
1. We sub out a similar Toshiba part. Luckily, Toshiba makes some similar parts, some of which are in stock at lots of places. How similar? They have exactly the same specs and use exactly the same charts and graphs as the parts we spec'd. Now, the question of why they'd produce several (seemingly) identical parts befuddles me, but if they are really that similar, they should drop in and work. I have some of these coming on Monday and will drop them in and re-qualify. If they work, then we're set. If they need additional compensation/parts tweaks, then it may cause a few days delay (if we have to go this way.)
2. We sub out a similar On Semiconductor part. On Semiconductor also makes similar parts. Not as similar as Toshiba, but they claim to have a drop-in replacement. This is a lot more iffy, because a different manufacturer's parts are, well, gonna behave differently. We'd have to take a long, hard look at them, even if they seemed to work. This might mean a longer delay. Or not. As with the Toshiba alts, I have some coming in Monday to see how they work.
On the plus side, we'll (at least) be getting data on a lot of alternate parts for Vidar (or other possible power amps). It certainly can't hurt in the future!
Stay updated on Schiit Audio at their sponsor profile on Head-Fi.
|