[sdiy] Nanosecond delays - the clean or dirty method?

Tom Wiltshire tom at electricdruid.net
Fri Feb 26 21:59:23 CET 2021


If it works, it works, and I don’t see any reason why it shouldn’t. Brian’s point about tolerances is a fair point - a little cap like that is probably +/-20%, so you’d better make sure it works one E6 down and one up just to be sure!

Ultimately, it depends how critical this this. If you were going to build several hundred and sell them to paying customers they’ll be mightily narked off if it doesn’t work reliably in 100% of cases, and so will you after you’ve dealt with those annoyed people, so you might make a different decision. If it’s a fix for one unit, you can afford to try it, tweak it, etc, and it never needs to work in any other unit.


> On 25 Feb 2021, at 17:39, Steve Lenham <steve at bendentech.co.uk> wrote:
> 
> Thanks, chaps. It's useful to get a second opinion.
> 
> I think I may struggle to get 5m of coax inside a RAM cart, so will probably stick with the little RC if I ever need to build one of these in anger (which I probably won't - it was just a little experiment to prove the concept at this stage).
> 
> Cheers,
> 
> Steve L.
> Benden Sound Technology
> 
> 
> On 25/02/2021 16:36, Richie Burnett wrote:
>> All seems perfectly reasonably to me Steve.
>> For smaller delays you can put little wiggles in the PCB trace or send the signal down a short length of transmission line. But you'd still need several metres of coax to nudge it back by just 25ns!
>> The slow rise time resulting from the RC circuit is not a problem in itself, but can cause problems in the presence of noise if the receiving device is fast enough to register the additional clock edges.
>> To be honest I've seen more problems with lightening-fast clock edges due to signal reflections. So I'd take a clock line with controlled dv/dt any day of the week :-)
>> -Richie,
>> Sent from my Xperia SP on O2
>> ---- Steve Lenham wrote ----
>>> I need to delay the Chip Enable signal to a RAM IC by a tiny amount -
>>> around 25ns. I'm building a replacement memory cart for a vintage unit,
>>> and this is necessary to allow the use of a battery-free F-RAM for
>>> storage (some F-RAMs have certain timing quirks).
>>> 
>>> I know that the proper way to do this would be to insert one or more
>>> pairs of inverters to delay the signal by a few propagation delays, but
>>> that needs, like, another chip.
>>> 
>>> The DIRTY way to do it is to add a tiny RC delay. I tried 1K5 and 22pF
>>> and it worked perfectly, delaying the edge by about 50ns and solving my
>>> problem.
>>> 
>>> I'm well aware of the dangers of slow pulse edges and would never
>>> consider this for something more lengthy without adding a Schmitt
>>> buffer, but am curious as to whether it is still seen as irredeemably
>>> bad practice for a tiny nudge such as this. I'm sure I have seen it done
>>> in various vintage schematics.
>>> 
>>> Any thoughts would be appeciated!
>>> 
>>> Cheers,
>>> 
>>> Steve L.
>>> Benden Sound Technology
>>> 
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at synth-diy.org
>>> http://synth-diy.org/mailman/listinfo/synth-diy
>>> Selling or trading? Use marketplace at synth-diy.org
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> Selling or trading? Use marketplace at synth-diy.org





More information about the Synth-diy mailing list