Parts delivered – check!
Assemblers ready – check!
PCBs tested – check!
PCBs tested? Are you sure? – Errrrr…
I was hoping to have had some units ready last week, although that would have meant everything would have gone smoothly.
Manufacturing is totally unlike my real job. In my day job, where I deliver and troubleshoot enterprise financial technology systems, time really is money, and when you say you’ll have something by a given time, in general you aim to do just that. If there are unforeseen glitches during the way, you flag them up at the time, not after your deadline when it’s all too late. Common sense you’d thing. Managing expectations is all.
Manufacturing is different, a complete mind shift. Over promising and under delivering seems to be the name of the game.
So, for this next batch of 500 units, we had a change to the PCB layout, predominently to integrate the HPF to save me days of monotonous of rework, but also I managed to squeeze on a bias tee. Two test panels (10 PCBs per panel) arrived from the PCB house (albeit a day late, but not a complete disaster) and built and tested fine. In the meantime, 52 more production panels were ordered from the same board house but on an 8 day turnaround. That turned out to be 11 days, due to St Patrick’s day, the moon phase and some other reasons seemingly too complex for me to understand, and equally apparently unnecessary to to tell the customer.
Once the production boards finally arrive, I take out a couple of panels for myself from the production bacth to check while the others are left with the assemblers. What can go wrong I ask myself?
Uh-oh, they have manufactured the production PCBs on 1.6mm FR4, not 1mm as specified, and as used in the test units, and as I flagged up in emails, and on the telephone when the original quote came through from the board house. So I am back to another “8 days”.
What’s wrong with 1.6mm boards?
The problem is that the SMA edge connectors that I use and have ordered and are in stock, ready, are for 1mm PCB. The PCB lands, the PCB width, the length of the threaded socket are all carefully chosed to be just right. Not all SMA edge connectors are made the same. Too short and you won’t be able to properly screw on a plug once in the enclosure. If the edge fingers are too long they’ll hit other parts on the board. Too wide and they won’t fit the PCB lands or the routed indent.
So I spent three days last week tracking down appropriate 1.6mm edge connectors that would be compatible. I ordered 550. They have already made the jouney from Hong Kong to Heathrow, then to East Midlands, then back to Heathrow, now they are at the Lambeth distribution centre. So tantalisingly close. Maybe I’ll get them today, maybe tomorrow. Until I have them and have tried them I’ll no idea as to whether they’re right or not.
So wish us luck, or we’ll have to hold on another week.
I used to be an RF design engineer working on the now defunct 2nd generation cordless phones, CT2 and in later years on the analogue parts of 10Gb/s optical fibre transmission equipment. We had many issues with the PCB boards not being of a consistant build up. It was common practice to reserve a small area separate from the main board as a test piece so we could measure track impedance before a batch of boards were populated. We used a multi layer board and many contolled impedance striplines throughout, so it was really important for the build up to be exactly as specified.
Will the change in board thickness cause any problems with controlled inpedance lines such as the one to the SMA connector and any bits around the front end on the FCD?
I wish you every success with the project, I am very pleased with the performance of mine.
The units on 1.6mm have no measureable difference in performance to those on 1mm.
Because it’s so tight in there there isn’t much in the way of any impedance hump to upset things: the distance from the SMA centre pad to the first 0402 component (not the land, the part itself) in the high pass filter is only 0.6mm.
Great job Howard!
I am looking forward to operate a FUNCube Dongle.
73, Gaston – ON4WF