Attiny25/45/85 FW Development Thread

Sorry, I edited my post above. For the ATTINY45:

^ That is what I'm using. I just push the pins in. They actually attach much better that way to programmer clips.

I’m assuming you mean the ATTINY45? I ordered a couple. Can you provide a pic of that (push the pins in)?

^ Yes, the 45. Here is one with the pins pushed about 1/2 way from stock position to all the way against the chip.

Here is one with the pins pushed all the way up against the chip.

Dang, that does look better for the SOIC clip! Might have to try one this eve - maybe this will solve the problems I'm having.

It looks like it's gonna walk away Smile

Yeah, it not only makes the chip taller, it increased the first curve of the pin which makes for a better "grab".

Just make sure you push on that inside arch near the end of the pin. Not on the actual end of the pin. In my first pic above, I pushed on the ends of the pins. That causes uneven leg lengths and the need for adjustments to get the pins to rest on the board evenly. You really only want to bend the pins once and then reflow to the board. After bending the pins, I wouldn't connect a programming clip until the chip is reflowed. The legs are much stronger soldered to the board. I may be being too cautious, but I don't like losing MCU's.

If that bothers you, imagine upgrading the hardware of a fleet of big rack-mounted cellular servers, each with 8 cells and 2592 pins per cell (20736 pins per server), requiring quite a bit of force to remove and insert each cell. Bent pins happened rather a lot. It’s fun trying to fix a single bent pin in the middle of a 48x54 grid deep inside a dark narrow cell slot. Especially when the cell guidance rails have poor tolerances due to being early prototypes, so where they land is somewhat random.

I’m glad I don’t work in a “big iron” data center any more.

^ That sounds crazy. That project would push me over the edge for sure.

BTW, Atmel was just bought out by Dailog Semiconductor (Dia who? Never heard of them… ):

Too slow Atmel acquired by Dialog

I wonder how lower frequencies are generated. Simple mechanisms like non-linearity only generate higher harmonics, overtones. Maybe the controller doesn’t keep the 19 KHz constant?

There has to be another frequency for that, maybe a loop in the code?

Skip the 45v and go straight to the 85v. It’s cheaper than the 45v on mouser and digikey, in single quantities and every other quantity level.

Thanks, will do. Need to order a Pomano 5250 clip also. Just broke mine in trying to get it to work. At least pretty sure I know wut the problem is now - too much solder on the legs, could be the over-used clip also. Think when the solder paste creeps up on the top of the legs, can't get a good grasp. I have a 25 based driver built up for a AS31 light sitting on the stock stripped driver with the switch and was able to get my old clip (black no name) working on it, so can hopefully get some testing done.

I got in my new Pomona 5250 clip now. I still can't reprogram a mounted 25V. I took an 85V, bent in the legs. I can "air" reprogram it over and over again. So I mounted it on a A22DD+1 board, continuity checked all the pins. Looks like a real nice setup for developing with - should be easy to get good contact on the clip. So, clipped it on - the test AVRDUDE command works every time - sees the ID as an 85. But when I program it, it downloads, and fails the verification of the program on the first byte every time - every time over and over. It sure looks like a good clip-on. With the bent legs, looks easier to get a good contact on all the pins. I tried several things - cleaning the pins, shimming it around - I don't know what else to do or try at this point -- out of ideas. Is there something on the driver board circuit maybe interfering with the 85?

This is the error:

avrdude: verifying ...
avrdude: verification error, first mismatch at byte 0x0000
0x4a != 0x00
avrdude: verification error; content mismatch

Any help or suggestions are extremely welcome...

Update: Just tried the "-V" option to turn off verification, but didn't help. It proceeded in programming the fuses but failed on each. The driver, also, doesn't seem to work at all.

Update: Also tried another stock 85V in "air" programming it -- works every time over and over again.

^

You know much more about this stuff than me, but I have to say that it does sound like something in the circuit design is interfering. Which board are you using and does it just have the typical components connected?

Edit: opps, the A22DD+

EDIT2: That is wight's 22mm board, right? I take it you don't have the OTC cap installed.

Right, no OTC cap. I'm using brownout detection for the power switch. I'm so close............

I think I have that driver with the 45V. I can try to flash it tonight and see what happens. Can you post or email me the exact FW you are trying to flash and the fuses?

I also have this one that is modded.

Ahhh - the firmware is my under test/debug/pre-cursor/not-quite-there version, if you know what I mean...

I'm looking around, and found someone else: vex_zg, who seems to have had similar experiences, here: https://budgetlightforum.com/t/-/31826.

So, he set 8 Mhz and turned off the /8 clock option... Hhmm - I'll try anything. He has very similar experiences though - the MCU seems bricked then comes back to life! Exactly what I saw. I'm also pm'ing RMM, TK, and Mike C and asking them to check in. I know Mike has done 85 work, and RMM and TK I think started.

Are you programming at 5v?
Slower avrdude programming speed? Like -B 4. Higher is slower, max is 64.