Use note on Tetroid buffered backplane...

Nagging hardware related question? Post here!
User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Use note on Tetroid buffered backplane...

Post by Dave »

I use the quite nice Tetroid backplane in my set-up, especially as I am regularly adding and removing cards.

I was testing a floppy interface and it failed after a few hours of use. It was the only card. The two failed 74-family logic ICs were replaced but it failed again shortly after, the same way. I went over the whole card, then measured it in the backplane. It turns out that the card was being damaged by reflections when used in the slot closest to the QL if nothing is in the other slots. If the card is moved to the end slot, the reflections are much smaller.

I recommend people should make sure to populate the expansion sockets from the left edge first to the QL last. If there's only one card, it should be in the left most slot.

This is not a knock on Tetroid's work. It's a perfectly fine backplane. Maybe my floppy interface is just sensitive. I was just a bit surprised by the reflections being so strong. +7.2V.


User avatar
NormanDunbar
Forum Moderator
Posts: 2251
Joined: Tue Dec 14, 2010 9:04 am
Location: Leeds, West Yorkshire, UK
Contact:

Re: Use note on Tetroid buffered backplane...

Post by NormanDunbar »

Hi Dave,

Can you explain "reflections" to me please? Consider my knowledge level akin to that of a 7 year old!

Thanks.
Norm.


Why do they put lightning conductors on churches?
Author of Arduino Software Internals
Author of Arduino Interrupts

No longer on Twitter, find me on https://mastodon.scot/@NormanDunbar.
User avatar
tofro
Font of All Knowledge
Posts: 2685
Joined: Sun Feb 13, 2011 10:53 pm
Location: SW Germany

Re: Use note on Tetroid buffered backplane...

Post by tofro »

NormanDunbar wrote:Hi Dave,

Can you explain "reflections" to me please? Consider my knowledge level akin to that of a 7 year old!

Thanks.
Norm.
Well, best explained with a water analogy:

Just like water in the seas that cannot instantly get to the places where it's "needed" (to equal out the water surface) and thus forms waves, the electrical signals in a wire cannot instantly get to their destination (they're limited by the speed of light) - so, a signal forms a wavefront, and a modulated (i.e., changing, like, for example, the 68008's clock) signal forms a lot of wave peaks and throughs on the conductor.

Next analogy: When water waves collide with solid surfaces (like a quay wall, for example) that are not prepared to eat up the energy of the wave, that energy must go somewhere - it is simply reflected and sent back to where it came.
The same happens with electrical signals at the open end of conductors - a wave comes in, the energy cannot go anywhere, so it's reflected and simply goes back.

And just like in water - when a reflected wave that goes back to its origin overlaps with a newly incoming one (that is, when the frequency of the original signal just coincidentally correlates with the run-time of the wave on the conductor), the energy of both wave peaks just adds up and can go even higher than the original signal (that's why you can, in such cases, measure more than 5 Volts on a TTL line). Such peaks cannot only disturb your electronics, the extreme "monster waves" can even easily kill them.

Reflections are normally suppressed by means of termination resistors to ground that act on electrical waves just like a nice, broad, sandy beach that eats up the energy of the water waves does, and no further reflection can occur. The same happens when you put a card on the extreme of the bus extender: Now the card and its ICs acts just like the termination resistor and eats the energy.

The whole phenomenom can also be intended. On your PC's WiFi-board, there's a piece of wire that has exactly the length of the wavelength of the signal that runs across it - your antenna - Because if you get the conditions just right, reflections on the conductor can actually leave the wire and go aereal (the analogon here is splashes and foam that are generated on heavy water waves) - But that's not what we want in a QL....
Last edited by tofro on Sat Apr 23, 2022 10:20 am, edited 2 times in total.


ʎɐqǝ ɯoɹɟ ǝq oʇ ƃuᴉoƃ ʇou sᴉ pɹɐoqʎǝʞ ʇxǝu ʎɯ 'ɹɐǝp ɥO
User avatar
Ruptor
Gold Card
Posts: 418
Joined: Fri Dec 20, 2019 2:23 pm
Location: London

Re: Use note on Tetroid buffered backplane...

Post by Ruptor »

Dave wrote:I use the quite nice Tetroid backplane in my set-up, especially as I am regularly adding and removing cards.
I was testing a floppy interface and it failed after a few hours of use. It was the only card. The two failed 74-family logic ICs were replaced but it failed again shortly after, the same way. I went over the whole card, then measured it in the backplane. It turns out that the card was being damaged by reflections when used in the slot closest to the QL if nothing is in the other slots. If the card is moved to the end slot, the reflections are much smaller.
Are there no terminating resistors on the backplane like they use on SCSI for instance? I wondered what the radio interference was with the transmissions coming from Texas due to the backplane tracks becoming aerials. :lol:


User avatar
NormanDunbar
Forum Moderator
Posts: 2251
Joined: Tue Dec 14, 2010 9:04 am
Location: Leeds, West Yorkshire, UK
Contact:

Re: Use note on Tetroid buffered backplane...

Post by NormanDunbar »

Thanks very much ToFro for that easy to understand explanation. I remember many years ago, fitting CB radios to boats, when I was a marine "engineer", and having to minimise the SWR once fitted with a (new) antenna. Sounds similar to what you were talking about.

Also, from secondary school physics, we have interference between waves -- some destructive (trough + crest, cancels out), or constructive (trough + trough = big trough, or crest + crest = big wave!) so I fully understand what you were saying about it.

Much obliged, thanks.

Cheers,
Norm.


Why do they put lightning conductors on churches?
Author of Arduino Software Internals
Author of Arduino Interrupts

No longer on Twitter, find me on https://mastodon.scot/@NormanDunbar.
User avatar
Pr0f
QL Wafer Drive
Posts: 1298
Joined: Thu Oct 12, 2017 9:54 am

Re: Use note on Tetroid buffered backplane...

Post by Pr0f »

I wonder if the tracks are wide enough to introduce some small SMD resistors in line of a few tens of ohms max? I seem to remember that there are some holes in one part of the 2pcb kit that might allow resistor packs too - will have to have a look when I get home.


User avatar
Pr0f
QL Wafer Drive
Posts: 1298
Joined: Thu Oct 12, 2017 9:54 am

Re: Use note on Tetroid buffered backplane...

Post by Pr0f »

Just found some pictures - so it looks like I imagined that.


User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Use note on Tetroid buffered backplane...

Post by Dave »

Looks like questions were asked and eloquently answered. :D

There's no need for termination on the backplane if a card occupies the last expansion slot.

If a backplane were to have termination, it can become quite a complex issue. Passive termination usually involves a resistor to ground, but that represents a load on the bus. To get enough termination, the resistance needs to be quite low and the load increases. Spread that across every line and the cumulative total can completely change the power requirements of the board. Active termination usually involves an external power supply that tries to weakly drive the lines from the end towards a neutral voltage. This has advantages of not being an apparent load to the logic, and working disproportionally on the most deviant signals -- so a full +5V or 0V are furthest from 2.5V and are most strongly controlled.

But yeah, just moving the card solves the problem too!


User avatar
Ruptor
Gold Card
Posts: 418
Joined: Fri Dec 20, 2019 2:23 pm
Location: London

Re: Use note on Tetroid buffered backplane...

Post by Ruptor »

Dave wrote:If a backplane were to have termination, it can become quite a complex issue. Passive termination usually involves a resistor to ground, but that represents a load on the bus. To get enough termination, the resistance needs to be quite low and the load increases.
I envisaged this kind of termination. They mention 68000 & other processors.
https://www.bourns.com/pdfs/rctermap.pdf


User avatar
Dave
SandySuperQDave
Posts: 2765
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Use note on Tetroid buffered backplane...

Post by Dave »

RC networks are fine if they're tuned to the characteristics of the circuit and the frequency of the rise time. The problem with the QL is that interfaces span enough period that the range of rise times/frequencies is quite broad, so you just have to target the center frequency.

When I say, "frequency" I'm not talking about the literal clock rate. I'm talking about the speed with which a signal transitions from 0 to 1 and back again. The speed at which that happens fits a notional sine wave of a higher frequency. In side the QL, some signals change quite slowly and have a frequency (for noise purposes) of around 20-22 MHz. Other areas have much higher speeds of voltage change that equal an idealized 60-65MHz. The video, memory, processing and microdrive sections have quite markedly different characteristics.

This is more of a problem because the power design of the QL is really, REALLY terrible. It's a two layer board with absolutely no consideration to return paths, which in the mid-80s was considered a state of the art consideration and had only really been applied in aerospace and communications up to that point. If Tetroid's QL board replacement was instead 4 layers, with the inner layers being unbroken power and ground fills, and all surface power and ground traces were eliminated, the QL board would be much more ideal. Not only that, removing the outer layer power and ground traces would allow many of the tracks to be repositioned more spaciously, grouped better, and eliminate a lot of crosstalk.


Post Reply