The Sinclair QL Forum

Hi Tom,

I wouldn't try surface mounting an ATmega328 myself to be honest, but people say it's easy, just wipe a load of solder on, then suck it back up with flux and braid. That's not a route I feel up to investigating, even with a microscope to check for micro shorts between pins!

I did mention that I was thinking of an DIP packaged ATmega328P rather than the surface mount variate -- although the latter would be more low-rise than the DIP, obviously.

Cheers,
Norm.

Derek_Stewart wrote:Would I be correct in thinking that the new internal Oqtadrive for MDV1, can function with an external device connected to the External Micdrove connector to use an Extetnal Oqtadrive or vDriveQL, ehich can access MDV3-MDV8?

Yes, that would work. You could use the internal h/w drive plus external ones. Only limitation would be that you have to map them as a group, e.g. 3,4,5 or 7,8 but not 2,5,8.

NormanDunbar wrote:I'm wondering, is the USB required after programming the Nano?

As Tom already said, in the combination with the Pi it's not needed, so de-soldering the USB socket would be a quick & easy option that would give a bit more head room.

NormanDunbar wrote:If not maybe the Nano mounting stuff could be replaced by the minimum required to keep an ATmega328P (DIP version) in operation ...

That would also be interesting, but I'd keep an external crystal and run the ATMega at 16MHz. I don't think it will work at 8MHz. Also, I'm not sure how stable the internal oscillator is. Exact timing is really crucial.

Alex

I'd like to add a note here from bitter personal experience: the internal oscillator on the ATmega328P is quite inaccurate. It can easily be +/- 5% or so. As this is a timing critical operation, I'd always use an accurate and more stable external clock source. For some reason, the internal oscillator in the DIP parts is less accurate than in the QFP parts. 5% is worst case - most are 2-3%. Also, the external oscillator can be 16 or 20MHz but the internal one is 8 MHz, with access to a /8 divider if you want a really low power system.

Fun fact: a 328P has slightly more processing power and memory than a PDP-11/03 mainframe.

I read the stuff on GitHub, it appears that the ATmega328 must run at 16 MHz as there are critical timings involved. I haven't read the source yet to see where, or if it could be adjusted though.

Dave, I'm sure there's an adjustment for the internal oscillator but I'd need to consult the data sheet to be sure. If I get a chance, I might configure a test device to output the clock on PORTB and check with my "oscilloscope".

Cheers,
Norm.

It's a common problem. It may not be the problem you're experiencing, but at least it's easy to check and eliminate.

Hi Dave,

I checked the data sheet. The internal 8MHz oscillator is factory "calibrated" to 8 MHz at VCC = 3V, Temperature = 25C with an accuracy of +/- 10%. It can be calibrated by the user to anything between 7.3 and 8.1 MHz with VCC between 1.8 and 5.5V, temperature from -40C to +85C and with an accuracy of +/- 1%.

There's no way my "scope" (Espotek Labrador) is accurate enough to measure.


Cheers,
Norm.

+/- 10% by spec is pretty terrible. Ooof. I'm glad their real world trimming is better than that.

I'll stick with using external oscillators, if only because fine tuning 500 boards to have accurate clocks when warmed up simply isn't practical. Also, many of my boards run at 20 MHz or 1 MHz.

NormanDunbar wrote:There's no way my "scope" (Espotek Labrador) is accurate enough to measure.

You need one of these.
https://www.ebay.co.uk/itm/274873141928 ... Swrthg9a-4

For those interested in building your own I've updated the PCB to v1.2. This basically shifts and rotates the Nano so it can be mounted in a socket rather than direct solder.

Shared project is on PCBWay https://www.pcbway.com/project/sharepro ... a1c12.html

I also have a few to sell as I got 10 in the order and only really need 2 or 3. £4 each if you want one in glorious black