This is something I want to test myself but can't yet and I'm curious if anyone else has tried it.

I mostly use all 74LVC ICs that are 3.3V TTL voltage levels; and with the exception of some slow peripheral devices the only 5V IC I use is the '816. I'm wondering if it is possible to run it at 20- to 25MHz at 3.3V * **.

Some of why I'm asking is just curiosity but some is because I've seen a couple of forum members using 3.3V ICs (mostly CPLDs) and I suspect the answer could be "run your 65x at 3.3V" rather than deal with voltage translation. I'm pretty certain that up to the official 14MHz the S6 fabricated ICs could easily manage at 3.3V (specifically the W65C02S6TQG-14 and the W65C816S6TQG-14). I mentioned here that the address out response time is almost constant all the way from 5V down to 4.2V. Unfortunately I can't test any lower than that until I have a new PCB manufactured.

Another thought I've had - to avoid voltage translation and potentially the time it takes - is to run the '816 at 3.8V to 4V which should interface with 74LVC ICs as their VOH tends to be quite close to 3.3V even when driving several other ICs.

* That would make my life a bit simpler, not much because it doesn't decrease my part count but, it does decrease my different types of part count.
** I'm not targeting 20MHz but I am (still) running an asymmetric clock with PHI2 low for 25ns and high for 75ns. That gives me so much more time to do funky things with the address.

I've done some frequency vs voltage characterization on CRC65 here.
viewtopic.php?f=6&t=6440&p=96894#p96894

Bottom line is W65C02 can run at 22MHz at 3.3V. Yes, it is much simpler to run W65C02 at 3.3V to interface with FPGA

Bill

well the only time i've run a 65816 at 3.3V was as an expansion module on my FPGA Dev Board.
and it runs stable at ~19MHz.

so with a regular SBC (one with actual ICs on a bus instead of everything going into an FPGA) at 3.3V i feel like you should atleast be able to reach 14MHz.
of course the main advantages of going 3.3V are being able to easily use FPGAs, Raspberry Picos, etc. without level shifting. and i think for something battery powered, going with lower voltages is a good idea. (would be interesting to see a very low power system running at 1.8V or something)

Thanks! I remember seeing - and then forgetting - you had posted that. It's good to know it will run up to 22MHz at 3.3V.
Thanks, that's good to know too!

I mentioned 25Mhz because I have a dream. Or at least I really, really want to run it at 12.5MHz. There's a couple of reasons why. Firstly it was my very originally specc'ed speed and I'm nothing if not a sucker for my own nostalgia. Mostly it's that the extra 12% performance just pushes the graphics blitter to the point where it can push out my (fairly arbitrary) required pixels per frame. And lastly it's because if I ever do get 4 '816s running in parallel then I can claim I have 50MHz of '816 *Electroboom voice*

The right thing for me to do is probably just to keep on running at 5V with voltage translation - the 74ALVC164245 has a propagation delay of about 4ns and the 74LVC4245 is about 3.5ns. But I'm still really curious to know if I could run the '816 at 3.8V and have it interpret TTL signals correctly. (because I have a whole bunch of 3.8V LDOs because reasons). The datasheet does not make me optimistic because VIH is given as VCC*0.8. Or 3.04V at 3.8V VCC. But to me WDCs datasheets seem to be conservative and it's very likely that the more 'standard' VCC*0.7 would work fine. That brings VIH to 2.66V at 3.8V VCC and that's well within the measured outputs of the LVC ICs I'm using.

If performance were the only goal then I'd call the above silly because dropping the '816 to 3.8V would probably slow it down more than the '4245 translation would. If simplicity is my goal (well I completely missed it already) but it does mean I can use the more common '245 instead; and not have to up translate other things like PHI2 and the interrupts.

A bit of a random question going the other way: has anyone seen if running at 6- or 7V makes a performance difference?

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Thanks, I'm assuming the the FCT245 will work be because even though their VIH is 2V its VOH is going to be close to 5V (assuming 5V VCC). The 241s are cool because I hadn't realised that there are buffers with opposite enables.

I also hadn't realised the LVC1Gs can happily run all the way up to 5V VCC - I first met them a year or two ago and got a slight surprise when I read the datasheet (ALWAYS READ THE DATASHEET. ALWAYS ). I just assumed they were the standard LVC voltages. They become a bit numerous / expensive for busses but I hadn't thought of using them for single trace signals like a clock or an interrupt before now.

I probably shouldn't mention 74LVCxxx seem to run stably for days at 4V. Probably fine for personal hobby projects but the current draw almost quadrupled (however I could not feel a temperature change).

I opted to change everything over to 3.3V last year some time. I started by building an extra PCB up using 3.3-volts parts. and swapped over to the older Microdrive with the 35-pin ZIF PATA connector. These are 3.3v only, and I'm using 3.3v PLD/CPLD parts going forward along with some 74LV parts.

Granted, I've only tested to 8MHz to date. I need to swap out the AT28BV256 EEPROM (200ns rated) with a faster NOR flash and then see where the next bottleneck is... likely the Microdrive as the ATA interface is really only spec'd for 8MHz.

_________________
Regards, KM
https://github.com/floobydust